I've been running Mozilla (SWB) V1.7-11 which I downloaded from HP for a
while.  It's been sufficient for my needs and even on quite old DEC
(yes, DEC, not even Compaq) hardware (EV56) it is fast enough for my 
needs.

A site I sometimes visit now crashes it.  Yes, there were changes to the 
site and yes, the programming of the page in question is probably not 
the best, but still, an application shouldn't crash even if fed invalid 
input.  The error is below.

Is there any "better" (read: more modern) web browser for VMS on ALPHA? 
It doesn't have to be from HP and I don't mind compiling and linking it, 
but it should build out of the box and not require a lot of effort on my 
part.  In order to navigate to some sites I need, it would have to do 
JavaScript.

---------8<-------------------------------------------------------------------

%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000000
000C, PC=00000000006AB7EC, PS=0000001B
%TRACE-F-TRACEBACK, symbolic stack dump follows
  image    module    routine             line      rel PC           abs PC
 LIBMOZJS  JSFUN  call_enumerate        20044 000000000000182C 00000000006AB7EC
 LIBMOZJS  JSFUN  js_PutCallObject      19869 00000000000012E4 00000000006AB2A4
 LIBMOZJS  JSINTERP  js_Invoke          20619 000000000000175C 00000000006B338C
 LIBMOZJS  JSINTERP  js_Interpret       22629 000000000000A718 00000000006BC348
 LIBMOZJS  JSINTERP  js_Invoke          20603 00000000000016F0 00000000006B3320
 LIBMOZJS  JSINTERP  js_Interpret       22629 000000000000A718 00000000006BC348
 LIBMOZJS  JSINTERP  js_Execute         20804 0000000000001EC8 00000000006B3AF8
 LIBMOZJS  JSAPI  JS_EvaluateUCScriptForPrincipals
                                        23874 0000000000008E48 0000000000680E48
 LIBGKLAYOUT  NSJSENVIRONMENT  EvaluateString
                                        66280 000000000000403C 00000000017F031C
 LIBGKLAYOUT  NSSCRIPTLOADER  EvaluateScript
                                        70000 00000000000046B4 0000000001515274
 LIBGKLAYOUT  NSSCRIPTLOADER  ProcessRequest
                                        69917 0000000000003FF0 0000000001514BB0
 LIBGKLAYOUT  NSSCRIPTLOADER  OnStreamComplete
                                        70238 00000000000059F4 00000000015165B4
 LIBNECKO  NSSTREAMLOADER  OnStopRequest
                                        51887 0000000000000F1C 0000000000F5195C
 LIBNECKO  NSHTTPCHANNEL  OnStopRequest
                                        74142 000000000001303C 0000000000FFA3BC
 LIBNECKO  NSINPUTSTREAMPUMP  OnStateStop
                                        47020 0000000000001C94 0000000000F2BA74
 LIBNECKO  NSINPUTSTREAMPUMP  OnInputStreamReady
                                        46861 0000000000001668 0000000000F2B448
 LIBXPCOM  NSSTREAMUTILS  EventHandler  14308 000000000000017C 0000000000AFBDFC
 LIBXPCOM  PLEVENT  PL_HandleEvent      41110 0000000000000E08 0000000000B21FE8
 LIBXPCOM  PLEVENT  PL_ProcessEventsBeforeID
                                        42137 0000000000001378 0000000000B22558
 LIBWIDGET_GTK  NSAPPSHELL  processQueue
                                        63377 0000000000001318 0000000001F593F8
 LIBXPCOM  NSVOIDARRAY  EnumerateForwards
                                        26036 00000000000010D4 0000000000ACC2A4
 LIBWIDGET_GTK  NSGTKEVENTHANDLER  handle_gdk_event
                                        69313 0000000000001BA8 0000000001F66A38
 LIBGDK  GDKEVENTS  gdk_event_dispatch  31328 0000000000002770 00000000001E9D10
 LIBGLIB  GMAIN  g_main_dispatch        19265 0000000000000B80 0000000000131FD0
 LIBGLIB  GMAIN  g_main_iterate         19486 000000000000132C 000000000013277C
 LIBGLIB  GMAIN  g_main_run             19544 0000000000001548 0000000000132998
 LIBGTK  GTKMAIN  gtk_main              21888 0000000000000AD8 00000000003BFDE8
 LIBWIDGET_GTK  NSAPPSHELL  Run         63262 0000000000000DC4 0000000001F58EA4
 MOZILLA-BIN  NSAPPRUNNER  main1        83970 0000000000006194 0000000000056194
 MOZILLA-BIN  NSAPPRUNNER  main         84450 0000000000007160 0000000000057160
 MOZILLA-BIN  NSAPPRUNNER  __MAIN           0 00000000000000B8 00000000000500B8
 MOZILLA-BIN                                0 0000000000072D64 0000000000072D64
 PTHREAD$RTL                                0 0000000000055E68 000000007BD33E68
 PTHREAD$RTL                                0 0000000000030404 000000007BD0E404
                                            0 FFFFFFFF80287F14 FFFFFFFF80287F14

In article <i87qoo$2ep$1@online.de>, helbig@astro.multiCLOTHESvax.de
(Phillip Helbig---undress to reply) writes: 

> Is there any "better" (read: more modern) web browser for VMS on ALPHA? 

On a related note, is there any way to get something like a flash player 
working as a browser plugin under VMS on ALPHA?

Assuming appropriate hardware, would it be possible to watch, say, a 
YouTube video, with sound, via a web browser running on ALPHA under VMS?

I don't mean "is it theoretically possible", but rather "where can I 
download the software" or at least "it would be possible if a skilled 
programmer puts in a few days' work".

Phillip Helbig---undress to reply wrote:
> In article <i87qoo$2ep$1@online.de>, helbig@astro.multiCLOTHESvax.de
> (Phillip Helbig---undress to reply) writes: 
> 
>> Is there any "better" (read: more modern) web browser for VMS on ALPHA? 
> 

Have you tried this one?

Secure Web Browser based on SeaMonkey v. 1.1.12

http://h71000.www7.hp.com/openvms/products/ips/cswbs/cswbs.html

> http://h71000.www7.hp.com/openvms/products/ips/cswbs/cswbs.html

And when you click on "Plug-ins" you can download the flash player 
plug-in. No sound, though, apparently.

The Flash plug-in for OpenVMS is provided by Mark Berryman, who ported 
Olivier Debon's open source flash player. The Flash plug-in for OpenVMS 
currently supports most of Flash 4 but does not support the actionscript 
features of Flash 5. *Support for sound is underway.* (emphasis added)

In article <i87rme$utn$1@speranza.aioe.org>, smithfarm
<presnypreklad@gmail.com> writes: 

> Phillip Helbig---undress to reply wrote:
> > In article <i87qoo$2ep$1@online.de>, helbig@astro.multiCLOTHESvax.de
> > (Phillip Helbig---undress to reply) writes: 
> > 
> >> Is there any "better" (read: more modern) web browser for VMS on ALPHA? 
> 
> Have you tried this one?
> 
> Secure Web Browser based on SeaMonkey v. 1.1.12
> 
> http://h71000.www7.hp.com/openvms/products/ips/cswbs/cswbs.html

I'll check it out.  This is confusing, though:

   Note: Secure Web Browser (based on SeaMonkey) and Secure Web Browser
   (based on Mozilla) are the same products supported on two different
   platforms. For more details, refer to the Release Notes.

Presumably I have (an earlier version of) the latter.

What are the "two different platforms"?

   Note: The build takes significant amount of time. For example, on a
   667Mhz DS20E AlphaStation, building a non-debug version of Mozilla
   takes about 12 hours. However, on a 900MHz, HP rx2600 Itanium, it
   takes about 10 hours.

Wow!

   In addition to the disk space requirements for the prerequisite
   software, the following is also required:
     * 500,000 blocks for the source code area
     * 1,200,000 blocks for a non-debug build area and/or 8,500,000
       blocks for a debug area

Wow!

   Linkname: Accept the license terms and download HP Secure Web Browser
          V1.1-12
   Filename:
          http://h71000.www7.hp.com/openvms/products/ips/cswbs/cswbs_register.html

Two questions: Does this link lead to the sources, or does it lead to
something installable (i.e. already built)?  If the latter, how much
disk space is needed a) for the (presumably compressed) download, b) for
the uncompressed download and c) during and after installation?  (I 
can't check yet since I might have to organise some disk space first.)

I also see some plugins.  Interesting.

In article <i87rsm$utn$2@speranza.aioe.org>, smithfarm
<presnypreklad@gmail.com> writes: 

> > http://h71000.www7.hp.com/openvms/products/ips/cswbs/cswbs.html
> 
> And when you click on "Plug-ins" you can download the flash player 
> plug-in. No sound, though, apparently.
> 
> The Flash plug-in for OpenVMS is provided by Mark Berryman, who ported 
> Olivier Debon's open source flash player. The Flash plug-in for OpenVMS 
> currently supports most of Flash 4 but does not support the actionscript 
> features of Flash 5. *Support for sound is underway.* (emphasis added)

Something to do during the winter holidays.  Probably a good idea to 
increase the power of my cluster and add some disk space first, though.

Phillip Helbig---undress to reply wrote:

>    Note: The build takes significant amount of time. For example, on a
>    667Mhz DS20E AlphaStation, building a non-debug version of Mozilla
>    takes about 12 hours. However, on a 900MHz, HP rx2600 Itanium, it
>    takes about 10 hours.



A real VMS engineer once posted something here about the build process
for this beast.  You need a whole slew of middleware apart from the
compilers so that the "make" can generate a bunch of .h and whatnot files .

Also, wasn't seamonkey compiled only on IA64, or did they eventually
release an alpha one ?

And there is always MOSAIC !

In article <4ca7937f$0$5897$c3e8da3$1cbc7475@news.astraweb.com>, JF
Mezei <jfmezei.spamnot@vaxination.ca> writes: 

> >    Note: The build takes significant amount of time. For example, on a
> >    667Mhz DS20E AlphaStation, building a non-debug version of Mozilla
> >    takes about 12 hours. However, on a 900MHz, HP rx2600 Itanium, it
> >    takes about 10 hours.
> 
> A real VMS engineer once posted something here about the build process
> for this beast.  You need a whole slew of middleware apart from the
> compilers so that the "make" can generate a bunch of .h and whatnot files .

Building it oneself seems a big undertaking.  Normally, I prefer to 
compile stuff myself, but with something like this, a pre-built version 
would be nice.  Maybe there is one.

> And there is always MOSAIC !

Probably not "modern" enough.  I often use LYNX, but if that fails, then
usually the big guns are required.  Yes, there is a niche where there
are graphical pages without bells and whistles, and Mosaic might be the
browser of choice there, but that means coordinating three browsers.  Of
course, Mosaic is probably the only reasonably modern alternative on
VAX. 

Phillip Helbig---undress to reply <helbig@astro.multiclothesvax.de> wrote:
> smithfarm <presnypreklad@gmail.com> writes:
>> Have you tried this one?
>>
>> Secure Web Browser based on SeaMonkey v. 1.1.12
>>
>> http://h71000.www7.hp.com/openvms/products/ips/cswbs/cswbs.html
>
> I'll check it out.  This is confusing, though:
>
>    Note: Secure Web Browser (based on SeaMonkey) and Secure Web Browser
>    (based on Mozilla) are the same products supported on two different
>    platforms. For more details, refer to the Release Notes.
>
> Presumably I have (an earlier version of) the latter.
>
> What are the "two different platforms"?

I'd say they are "Mozilla" and "SeaMonkey".

[...]
>    Linkname: Accept the license terms and download HP Secure Web Browser
>           V1.1-12
>    Filename:
>           http://h71000.www7.hp.com/openvms/products/ips/cswbs/cswbs_register.html
>
> Two questions: Does this link lead to the sources, or does it lead to
> something installable (i.e. already built)?

The latter.

> If the latter, how much
> disk space is needed a) for the (presumably compressed) download, b) for
> the uncompressed download and c) during and after installation?  (I
> can't check yet since I might have to organise some disk space first.)

See the Release Notes for Alpha
(http://h71000.www7.hp.com/openvms/products/ips/cswbs/cswbs_relnotes_alpha.html):

"The HP Secure Web Browser for OpenVMS Alpha compressed self-extracting
archive contains 80,511 blocks. The expanded PCSI kit requires an additional
150,000 blocks of working disk space to install."

cu,
   Martin
-- 
One OS to rule them all       | Martin Vorlaender  |  OpenVMS rules!
One OS to find them           | work: mv@pdv-systeme.de
One OS to bring them all      |   http://vms.pdv-systeme.de/users/martinv/
And in the Darkness bind them.| home: martin.vorlaender@t-online.de

In article <op.vjzdy3ozdc3j58@notemv-tap.mv.privat>, "Martin Vorlaender"
<mv@pdv-systeme.de> writes: 

> >> Have you tried this one?
> >>
> >> Secure Web Browser based on SeaMonkey v. 1.1.12

As I said, I now have Mozilla (SWB) V1.7-11.  I assume that 1.1.12 
SeaMonkey is NEWER than 1.7-11 SWB.  It looks like I installed Mozilla 
on 9-APR-2006.  They seem quite similar.  Presumably I can just install 
SeaMonkey on top of Mozilla (both are called CSWB) and it will overwrite 
the old version.

> "The HP Secure Web Browser for OpenVMS Alpha compressed self-extracting
> archive contains 80,511 blocks. The expanded PCSI kit requires an additional
> 150,000 blocks of working disk space to install."

OK, I have enough space, but will upgrade the cluster first! 

> As I said, I now have Mozilla (SWB) V1.7-11.  I assume that 1.1.12 
> SeaMonkey is NEWER than 1.7-11 SWB.  It looks like I installed Mozilla 
> on 9-APR-2006.  They seem quite similar.  Presumably I can just install 
> SeaMonkey on top of Mozilla (both are called CSWB) and it will overwrite 
> the old version.

SeaMonkey is the "continuation" of Mozilla Application Suite.

 From SeaMonkey Wikipedia article: "The development of SeaMonkey is 
community-driven, in contrast to the Mozilla Application Suite, which 
until its last released version (1.7.13) was governed by the Mozilla 
foundation. The new project-leading group is the SeaMonkey Council."

HTH
Nathan

In article <i89fsn$8ff$1@speranza.aioe.org>, smithfarm
<presnypreklad@gmail.com> writes: 

> > As I said, I now have Mozilla (SWB) V1.7-11.  I assume that 1.1.12 
> > SeaMonkey is NEWER than 1.7-11 SWB.  It looks like I installed Mozilla 
> > on 9-APR-2006.  They seem quite similar.  Presumably I can just install 
> > SeaMonkey on top of Mozilla (both are called CSWB) and it will overwrite 
> > the old version.
> 
> SeaMonkey is the "continuation" of Mozilla Application Suite.

OK, I guess there was a reboot of the version numbers at some point.

As long as the SeaMonkey at HP is substantially more recent than 2006, 
it is worth a try.

Hi Philip,

Ok, I tried not to bite but couldn't help it; why would anyone voluntarily 
be using a VMS-based solution to browse the web?

Absolute-fucking-disbelief!

Cheers Richard Maher 

> As long as the SeaMonkey at HP is substantially more recent than 2006, 
> it is worth a try.
> 

According to the release history at the Seamonkey Wikipedia article,
v.1.1.12 was released on September 23, 2008.

In article <i89jit$iau$1@speranza.aioe.org>, "Richard Maher"
<maher_rj@hotspamnotmail.com> writes:
> Hi Philip,
> 
> Ok, I tried not to bite but couldn't help it; why would anyone voluntarily 
> be using a VMS-based solution to browse the web?

maybe because a true VMS bigot would like to use 
his favourite OS for just everything,
including everyday tasks like browsing and mailing?

Or, turned the other way, why would anyone voluntarily
use a second OS for that?

On 10/3/2010 5:49 AM, Richard Maher wrote:
> Hi Philip,
>
> Ok, I tried not to bite but couldn't help it; why would anyone voluntarily
> be using a VMS-based solution to browse the web?
>
> Absolute-fucking-disbelief!
>
> Cheers Richard Maher
>
>

Why not?  Windows is certainly the most prevalent O/S these days but 
it's hardly the only O/S with the capability.  VMS is a lot less likely 
to become infected with an "electronic organism"!

In article <i89jit$iau$1@speranza.aioe.org>, "Richard Maher"
<maher_rj@hotspamnotmail.com> writes: 

> Ok, I tried not to bite but couldn't help it; 

You have hereby outed yourself as a soldier of the Dark Side.  :-)

> why would anyone voluntarily 
> be using a VMS-based solution to browse the web?

Why not?

It is MUCH easier to have just ONE system to maintain---VMS---than 
several different systems.  Right tool for the job, sure, but one of the 
strengths of VMS is the "desktop-to-datacenter" concept: the same 
software runs on the desktop as on the big iron.  If I download stuff, 
then it is stuff I want on VMS, so easier to just download it to the 
same computer the browser is running on.  And I have the advantages of 
VMS; in particular, I don't have to protect my systems from being 
converted into zombies by script-kiddies.

As long as there is a good web browser for VMS, why not?  Why use 
something else?  

In article <i89she$hc2$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de
(Michael Kraemer) writes: 

> In article <i89jit$iau$1@speranza.aioe.org>, "Richard Maher"
> <maher_rj@hotspamnotmail.com> writes:
> > Hi Philip,
> > 
> > Ok, I tried not to bite but couldn't help it; why would anyone voluntarily 
> > be using a VMS-based solution to browse the web?
> 
> maybe because a true VMS bigot 

I prefer the term "enthusiast".  :-)

> would like to use 
> his favourite OS for just everything,
> including everyday tasks like browsing and mailing?

Of course.  I've done all my mail using VMS mail since 1992 or 
thereabouts.  How quickly can other folks dig up an old mail from 15 
years ago to check something out?

Phillip Helbig---undress to reply mentioned  on 3-10-2010 21:02:
> In article<i89she$hc2$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de
> (Michael Kraemer) writes:
>
>> In article<i89jit$iau$1@speranza.aioe.org>, "Richard Maher"
>> <maher_rj@hotspamnotmail.com>  writes:
>>> Hi Philip,
>>>
>>> Ok, I tried not to bite but couldn't help it; why would anyone voluntarily
>>> be using a VMS-based solution to browse the web?
>>
>> maybe because a true VMS bigot
>
> I prefer the term "enthusiast".  :-)
>
>> would like to use
>> his favourite OS for just everything,
>> including everyday tasks like browsing and mailing?
>
> Of course.  I've done all my mail using VMS mail since 1992 or
> thereabouts.  How quickly can other folks dig up an old mail from 15
> years ago to check something out?

For the record, my personal e-mail system (switched to and from Msoft 
Outlook Express, Outlook, Netscape Mail, Eudora, and now Thunderbird) 
also contains mails from the early nineties. The mail folder structure 
survived all migrations. I used VMSmail and ALL-IN-1 when I was employed 
by Digital, but strictly as a business mail system, so I left that 
behind when I quit DEC in 1997.

/Wilm

Why run a browser on VMS ?

When you need to download stuff to your VMS system, having a browser on
it is convenient.

When you want to consult documentation on-line, having a browser is
convenient so you can cut/paste ode examples from the browser into TPU.

In article <4ca8dd1a$0$8922$703f8584@textnews.kpn.nl>, Wilm Boerhout
<w6.boerhout@planet.nl> writes: 

> > Of course.  I've done all my mail using VMS mail since 1992 or
> > thereabouts.  How quickly can other folks dig up an old mail from 15
> > years ago to check something out?
> 
> For the record, my personal e-mail system (switched to and from Msoft 
> Outlook Express, Outlook, Netscape Mail, Eudora, and now Thunderbird) 
> also contains mails from the early nineties. The mail folder structure 
> survived all migrations. I used VMSmail and ALL-IN-1 when I was employed 
> by Digital, but strictly as a business mail system, so I left that 
> behind when I quit DEC in 1997.

For those who want to go (back) to VMS mail, this blast from the past 
provides a way!  (I hope Elliott still follows the group.  There might 
be some undocumented TECO code below which will start up a lawnmower.)

---------8<-------------------------------------------------------------------

>Newsgroups: comp.os.vms
>Path: news.rrz.uni-hamburg.de!news.dkrz.de!dscomsa.desy.de!news.belwue.de!scsing.switch.ch!newsstand.cit.cornell.edu!news.kei.com!world!mv!mv.mv.com!morris
>From: morris@mv.mv.com (Skipper W. Morris)
>Subject: Tool to convert U**x mail to VMSMail
>Message-ID: <DK8E95.Dt1@mv.mv.com>
>Organization: MV Communications, Inc.
>Date: Wed, 27 Dec 1995 06:05:28 GMT
>X-Nntp-Posting-Host: mv.mv.com
>Lines: 241

VMS Mail lovers,
        I worked up the little gem below after another frustrating day of
trying to find a reasonable way to deal with lots of mail.  It converts U**X
mail file folders to VMS mail files/folders/messages.
        Bug reports/fixes/comments/etc, are appreciated.
                                        enjoy
                                                /Skip Morris
        
----------------------------------cut here------------------------------------
$ set noon						! 'f$verify(0)'
$!+
$! CONVERT_MAIL.COM V1.0, Skip Morris, "morris@mv.mv.com", 26-Dec-95
$!    Input paramters:
$!	P1 = The file(s) you want to convert.
$!	P2 = Output VMS Mail file.
$!	P3 = VMS Mail folder to store mail (defaults to input file name).
$!      P4 = if nonnull then only convert to Mail11 V3.
$!    Wildcard directory/filename inputs are permitted.
$!    Use from batch by:
$!      "$ submit convert_mail /param="DISK$FOO:[SOURCE...]*.txt;"
$!    Max supported message count per input file is 16,777,215 since
$!      that's all that could fit in the available return status fields.
$!    Any '"' char's in personal name are blown away as VMSMail barfs on 'em.
$!    Any special char's in addresses string cause address to be quoted.
$!    Warning!! properly formatted unix mail file assumed. Not much checking.
$!-
$ default = f$envi("procedure")
$ if f$parse(p1,,,"name") + f$parse(p1,,,"type") + f$parse(p1,,,"version") -
    .eqs. ".;" then p1 = p1 + "*.*;" ! cause directory input to be wildcard
$ search_file = f$parse(p1)
$ wildcard = f$loca("*",search_file) .ne. f$leng(search_file) .or. -
    f$loca("%",search_file) .ne. f$leng(search_file) .or. -
    f$loca("...",search_file) .ne. f$leng(search_file)
$ output_file = f$parse(p2,"mail.mai;",p1)
$ temp_file = f$parse("mail_"+f$getjpi("","pid")+".tmp;",output_file,default)
$ v3only = p4 .nes. ""
$main_loop:
$ input_file = f$search(search_file)
$ if input_file .nes. ""
$  then
$   write sys$output "Converting ",input_file," to Maill11V3 format..."
$   folder = f$parse(input_file,,,"name")
$   call convert_unix_mail 'input_file' 'temp_file'
$     ! return status has line count encoded in return status value
$   msg_count = ($status .and. %x07fffff8) / 8
$   if .not. v3only
$    then
$     write sys$output f$fao("Converting to ISAM file !AS folder !AS...", -
        output_file,folder)
$     call convert_vms_mail 'temp_file' 'output_file' 'folder'
$     delete 'temp_file'
$    else
$     write sys$output f$fao("New file is !AS.",output_file)
$     rename 'temp_file' 'output_file'
$    endif
$   write sys$output f$fao("...done, !ZL message!%S processed.",msg_count) 
$   if wildcard then goto main_loop
$  endif
$ exit
$!
$CONVERT_VMS_MAIL: SUBROUTINE
$ tempcom = f$parse("mail-"+f$getjpi("","pid")+".tmp;",p2,f$envi("procedure"))
$ create 'tempcom'
$ open/append file 'tempcom'
$ write file "$ mail"
$ write file "set file ",p1
$ write file "copy/all ",p3," ",p2,"/noconfirm"
$ write file "exit"
$ close file
$ @'tempcom'
$ delete 'tempcom'
$ exit
$ENDSUBROUTINE
$!
$CONVERT_UNIX_MAIL: SUBROUTINE
$ teco = "$teco32 teco"
$ unix_mail_file = f$search(f$parse(p1))
$   if unix_mail_file .eqs. "" then exit 4
$ temp_file = f$parse(p2)
$   if temp_file .eqs. "" then exit 4
$!
$ if f$trnlnm("tec$memory") .nes. "" then deassign tec$memory
$ assign/user 'unix_mail_file' mail$in
$ assign/user 'temp_file' mail$out
$ assign/user _nla0: sys$output ! comment this line out to vfy teco execution
$ define/user tec$init "96ee"
$ teco
! goto top, case match exact, load <cr><lf> qu, qt tokens illegal Mail-11!
j1^x13^uu`10:^uu`^ut-`              ! Add illegal in "From:" after "-" char!
9^uv`32:^uv`13:^uv`:^uv.:%@!^=/[]`  ! qv contains sendmail delimiter tokens!

ermail$in`ewmail$out`               ! open input, output files!

!main loop for each msg!
<%q                                 ! start loop, increment msg counter!
  b,.pb,.ka                         !   page out previous, get new!
  12i`                              !   insert <ff> at msg start!
  .ua                               !   store msg top in qa!
  s^el^equ`                         !   srch 1 blank lines!
  .ub                               !   store end of headers qb!
  qajl2r                            !   go end of initial "From" line!
  .-4,.xm                           !   get year in qm!
  5r.uc                             !   qc at end of timezone/timeofday!
  r-s `::s^ea`"s2r.ucr-s `'         !   if alpha timezone then backup!
  .,qcxn                            !   get time in qn!
  .-1uc-2s `.,qcxo                  !   1-2 digit day-of-month in qo!
  3r-s `                            !   go to start of month!
  .,.+3xp                           !   get month in qp!
  qaj                               !   search "From: " header!
  qb-qa,:s^elFrom: `"s              !   if successful!
    -d9i`                           !   then change space to tab!
    0l1xlkqajgl-l                   !     move line to top!
    6c.uc                           !     save start of address qc!
    0,0xc0,0xl                      !     init qc(pers nam), ql(addr)!
    0a-34"e                         !     if 'From: "pp" <aa>' sntx!
      c1:fb" `"s                    !     then if ending '" '!
        1:fb<`"s                    !       then if find starting '<'!
          1:fb>`"s                  !         then if ending '>' found!
            qcj                     !           then go start pers name!
            dqb-1ub                 !             delete leading '"', fixup qb!
            s"`-dqb-1ub             !             find ending '"', del, fix qb!
            qc,.xc                  !             store pers name qc!
            qc,.k                   !             kill pers name!
            dqb-1ub                 !             delete space end pers name!
            s<`-dqb-1ub.ud          !             find starting '<' and delete!
            s>`-dqb-1ub             !             find ending '>' and delete!
            qd,.xlqd,.k             !             store addr in ql and kill!
          |46^t.:=69^t1:=77^tqq=0tt !           else bad msg fmt, out err msg!
          '                         !           endif!
        '                           !         endif!
      | 1:fb"`"s                    !       else if 'From: "aa"aa (pp)' syntax!
          1:fb (`"s                 !         then if ' ('!
            1:fb)`"s                !           then if ending ')' found!
              qcj                   !             then go start addr!
              s (`-2dqb-2ub         !               delete ' (', fix qb!
              qc,.xlqc,.k           !               put addr ql, delete!
              s)`-dqb-1ud           !               go end pers, del ')', fx qb!
              qc,.xcqc,.k           !               put pers name qc, delete!
            |46^t.:=69^t2:=77^tqq=0tt !           else bad msg fmt, out err msg!
            '                       !             endif!
          |qcjl2rqc,.xlqc,.k0,0xc   !           else assume 'From: "aa"aa' sntx!
          '                         !           endif!
        |46^t.:=69^t3:=77^tqq=0tt   !         else bad msg fmt, output err msg!
        '                           !         endif!
      '                             !       endif!
    | 1:fb <`"s                     !     else if 'From: pp <aa>' sntx, if ' <'!
        1:fb>`"s                    !       then if ending '>' found!
          qcj                       !         then go start pers name!
          s <`-2dqb-2ub             !           find starting ' <' and delete!
          qc,.xc                    !           store pers name qc!
          qc,.k                     !           kill pers name!
          s>`-dqb-1ub               !           find ending '>' and delete!
          qc,.xlqc,.k               !           store addr in ql and kill!
        |46^t.:=69^t4:=77^tqq=0tt   !         else bad msg fmt, output err msg!
        '                           !         endif!
      |                             !       else!
        1:fb (`"s                   !         if 'From: aa (pp) sntx, if ' ('!
          1:fb)`"s                  !         then if ending ')' found!
            qcj                     !           then go start addr!
            s (`-2dqb-2ub           !             delete ' (', fix qb!
            qc,.xlqc,.k             !             put addr ql, delete!
            s)`-dqb-1ud             !             go end pers, del ')', fix qb!
            qc,.xcqc,.k             !             put pers name qc, delete!
          |46^t.:=69^t5:=77^tqq=0tt !           else bad msg fmt, outpt err msg!
          '                         !           endif!
        |qcj                        !         else assume single addr case!
          ::s<`"s-ds>`-dqb-2ub'     !           handle case 'From: <aa>'!
            l2rqc,.xlqc,.k0,0xc     !           assume 'From: aa' syntax!
        '                           !         endif!
      '                             !       endif!
    '                               !     endif!
    qcjgl.ud                        !     go end "From:", insert addr!
    qcj<qd-.,s^egt`;                !     loop find illegal chars addr!
      -s^egv`i"`s^egt`              !       quote previous delimiter token!
      s^egv`ri"`qb+2ubqd+2ud        !       quote next sendmail delimiter token!
      >                             !       end loop!
    :qc"g                           !     if pers name nonnull!
      qdji `qb+1ub                  !     then insrt space delimiter!
      i"`gci"`qb+2ub                !       insrt pers nam w/quotes!
    '                               !     endif!
  |qajiFrom:`9i`                    !   else insrt "From:" anyway!
    i"<unknown>"`13i`10i`qb+19ub    !     create "<unknown>" from, fixup qb!
  '                                 !   endif!
  qajqb-qa,:s^elTo: `"s             !   if 'To:' header found!
    -d9i`                           !   then chg space to tab!
    l::s `"s-3dqb-3ub               !     if additional 'To:' header lines!
      <::s `;-dqb-1ub>i `qb+1ub     !     then append & kill leading spaces!
      l::fs   `To:`                 !       Add 'To:' hdr if more addressee!
    '                               !     endif!
    -l1xlkqajlgl                    !     move 'To:' line to new loc!
  |qajliTo:`9i`i`13i`10i`qb+6ub     !   else insert empty 'To:'!
  '                                 !   endif!

  qajqb-qa,:s^elCc: `"s             !   if 'Cc:' header found!
    -3diC:`9i`                      !   then do same stuff!
    l::s `"s-3dqb-3ub               !     if additional 'Cc:' header lines!
      <::s `;-dqb-1ub>i `qb+1ub     !     then append & kill leading spaces!
      l::fs   `Cc:`                 !       Add 'Cc:' hdr if more addressee!
    '                               !     endif!
    -l1xlkqaj2lgl
  |qaj2liCC:`9i`i`13i`10i`qb+6ub    !   else insert empty 'CC:'!
  '                                 !   endif!

  qajqb-qa,:s^elSubject: `"s        !   if 'Subject:' found!
    -5di:`9i`qb-3ub                 !   then do it again!
    <::s `"u0;'-dqb-1ub>            !     delete leading spaces in Subj line!
    0l1xlkqaj3lgl                   !     move Subj line to Mail-11 header!
  |qaj3liSubj:`9i`i`13i`10i`qb+8ub  !   else!
  '                                 !   endif!

  13i`10i`qb+1ub                    ! mark end of Mail-11 hdr!
  qajl2r                            ! goto end of From: line!
  .uc                               ! mark current pos in qc!
  i  `goi-`gp                       ! insert timestmp dlmtr,yy,mm!
  3r0a&223i`d0a&223i`d0a&223i`d     ! upcase month!
  i-`gmi `gn                        ! add rest of timestamp!
  .-qc+qbub                         ! fixup qb!
  qbj-l                             ! goto end of headers!

  <n^el^equ`;                       ! loop to find blank line!
    .-z"ea'::sFrom `"s0l0;'         !   if endbuff append, if newmsg exit loop!
    ::s>From `"sfrFrom `|-l'>       !   replace '>From ' hack, next loop!
  ::sFrom `;0l                      ! exit loop if no more messages!
>``                                 ! end main loop!

!erase buffer, load local DCL symbol "c_line_count" with line
  count by: creating appropriate teco command, loading it in Q
  register A, and executing it!
ji:egsym msg_count `qq\27i`b,.xab,.kma``

ex``                                ! exit with status!
$!
$! exit and (hack alert) pass back count as part of return status in the
$! facility number and message number fields (set the suppress msg, customer
$! specific, and success status bits just to be politically correct).
$ if f$inte(msg_count) .le. %x07fffff8/8		! 16777215 in decimal
$  then exit (f$inte(msg_count)*8).or.%x18000003
$  else exit 1						! pass back count = 0
$  endif
$ENDSUBROUTINE

On 2010-10-03 22:21, JF Mezei wrote:
> Why run a browser on VMS ?
>
> When you need to download stuff to your VMS system, having a browser on
> it is convenient.
>

Much easier to use some tool like HTTP_FETCH or similar then to
run a full browser. And it's just as easy to keep SW-kits on my
laptop ready to be used at any particular site I'm working on.
Much faster to make a simple FTP copy then trying to get a
WEB browser running (on VMS).

> When you want to consult documentation on-line, having a browser is
> convenient so you can cut/paste ode examples from the browser into TPU.

I have absolutely no problem att all cut-n-pasting between by Windows
browser and my Reflection session ! That was realy dumb...

There is of course absolutely no *professional* reason at all to
fight the war against browsers under VMS and there are a heap
of reasons to run it on (some of the) most used desktop systems.

Now, if you add in historical, hobbyist or similar factors into
the pot, I don't know. But what has that to do with anything real ?

Trying to use your VMS systems for WEB access *professionally*
is just, well, unprofessional.

This is one of the things I don't know why HP are putting effort
into, WEB browsers for VMS. Totaly wasted money...

> For those who want to go (back) to VMS mail, this blast from the past 
> provides a way!  (I hope Elliott still follows the group.  There might 
> be some undocumented TECO code below which will start up a lawnmower.)
> 

Does it handle quoted-pukeable? Binary attachments?

On 03-10-2010 08:22, Michael Kraemer wrote:
> maybe because a true VMS bigot would like to use
> his favourite OS for just everything,
> including everyday tasks like browsing and mailing?
>
> Or, turned the other way, why would anyone voluntarily
> use a second OS for that?

Because the software are better on that OS?

Arne

On 03-10-2010 15:02, Phillip Helbig---undress to reply wrote:
> Of course.  I've done all my mail using VMS mail since 1992 or
> thereabouts.  How quickly can other folks dig up an old mail from 15
> years ago to check something out?

A lot easier than on VMS I would assume due to the sort
and search capabilities in newer email programs.

Arne

In article <i8ajvi$isu$2@online.de>,
	helbig@astro.multiCLOTHESvax.de (Phillip Helbig---undress to reply) writes:
> In article <i89she$hc2$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de
> (Michael Kraemer) writes: 
> 
>> In article <i89jit$iau$1@speranza.aioe.org>, "Richard Maher"
>> <maher_rj@hotspamnotmail.com> writes:
>> > Hi Philip,
>> > 
>> > Ok, I tried not to bite but couldn't help it; why would anyone voluntarily 
>> > be using a VMS-based solution to browse the web?
>> 
>> maybe because a true VMS bigot 
> 
> I prefer the term "enthusiast".  :-)
> 
>> would like to use 
>> his favourite OS for just everything,
>> including everyday tasks like browsing and mailing?
> 
> Of course.  I've done all my mail using VMS mail since 1992 or 
> thereabouts.  How quickly can other folks dig up an old mail from 15 
> years ago to check something out?
> 

I can go back 22 years.  That's when I first go there.  I can do it
in realtime.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

On 03-10-2010 15:00, Phillip Helbig---undress to reply wrote:
> In article<i89jit$iau$1@speranza.aioe.org>, "Richard Maher"
>> why would anyone voluntarily
>> be using a VMS-based solution to browse the web?
>
> Why not?
>
> It is MUCH easier to have just ONE system to maintain---VMS---than
> several different systems.  Right tool for the job, sure, but one of the
> strengths of VMS is the "desktop-to-datacenter" concept: the same
> software runs on the desktop as on the big iron.

If you could get FF 3.x and TB 3.x for VMS and an I64 box was
as cheap as a x86-64 box, then it would be nice to use VMS
for it.

But ...

Arne

In article <4ca8f5e0$0$50449$14726298@news.sunsite.dk>,
=?ISO-8859-1?Q?Arne_Vajh=F8j?= <arne@vajhoej.dk> writes:
> On 03-10-2010 08:22, Michael Kraemer wrote:
> > maybe because a true VMS bigot would like to use
> > his favourite OS for just everything,
> > including everyday tasks like browsing and mailing?
> >
> > Or, turned the other way, why would anyone voluntarily
> > use a second OS for that?
> 
> Because the software are better on that OS?
> 

Define "better".
If the browsers are the same, what's the difference?
Moreover, if (!) VMS is so invulnerable to virii etc
and allegedly unhackable, it should
be the best software for mailing and web-browsing, no?

In article <i8asds$f1b$1@news.albasani.net>, Jan-Erik Soderholm
<jan-erik.soderholm@telia.com> writes:

> 
> Much easier to use some tool like HTTP_FETCH or similar then to
> run a full browser.
> And it's just as easy to keep SW-kits on my
> laptop ready to be used at any particular site I'm working on.
> Much faster to make a simple FTP copy then trying to get a
> WEB browser running (on VMS).

Provided a simple ftp access is still offered.
With all those endless "click and confirm" interfaces
so much in fashion these days, automated tools are becoming
increasingly worthless.


> > When you want to consult documentation on-line, having a browser is
> > convenient so you can cut/paste ode examples from the browser into TPU.
> 
> I have absolutely no problem att all cut-n-pasting between by Windows
> browser and my Reflection session ! That was realy dumb...

but maybe people out there do not *want* windows ...
 
> There is of course absolutely no *professional* reason at all to
> fight the war against browsers under VMS and there are a heap
> of reasons to run it on (some of the) most used desktop systems.
> 
> Now, if you add in historical, hobbyist or similar factors into
> the pot, I don't know. But what has that to do with anything real ?
> 
> Trying to use your VMS systems for WEB access *professionally*
> is just, well, unprofessional.

calling other people's use of VMS unprofessional
without having walked in their shoes for one day is, well,
unprofessional.
 
> This is one of the things I don't know why HP are putting effort
> into, WEB browsers for VMS. Totaly wasted money...

Of course not. Keeps the OS viable.
I wouldn't want to use an OS which isn't capable 
to read it's own HTML docs.

smithfarm wrote:

> Does it handle quoted-pukeable? Binary attachments?



BREAKING NEWS:

smithfarm has been discovered to be VAXman's long lost twin brother !

Sweet commentary!
________________________________________
From: info-vax-bounces@rbnsn.com [info-vax-bounces@rbnsn.com] On Behalf Of =
JF Mezei [jfmezei.spamnot@vaxination.ca]
Sent: Sunday, October 03, 2010 5:47 PM
To: info-vax@rbnsn.com
Subject: Re: [Info-vax] Horses & water

smithfarm wrote:

> Does it handle quoted-pukeable? Binary attachments?



BREAKING NEWS:

smithfarm has been discovered to be VAXman's long lost twin brother !
_______________________________________________
Info-vax mailing list
Info-vax@rbnsn.com
http://rbnsn.com/mailman/listinfo/info-vax_rbnsn.com

On 03-10-2010 18:15, Michael Kraemer wrote:
> In article<4ca8f5e0$0$50449$14726298@news.sunsite.dk>,
> =?ISO-8859-1?Q?Arne_Vajh=F8j?=<arne@vajhoej.dk>  writes:
>> On 03-10-2010 08:22, Michael Kraemer wrote:
>>> maybe because a true VMS bigot would like to use
>>> his favourite OS for just everything,
>>> including everyday tasks like browsing and mailing?
>>>
>>> Or, turned the other way, why would anyone voluntarily
>>> use a second OS for that?
>>
>> Because the software are better on that OS?
>
> Define "better".

More things work.

CSS3, HTML5, Flash, SilverLigth etc..

> If the browsers are the same, what's the difference?

None.

But given that the browsers and plugins are not the same,
then that is not so relevant.

Arne

Hi Jan-Erik,

"Jan-Erik Soderholm" <jan-erik.soderholm@telia.com> wrote in message 
news:i8asds$f1b$1@news.albasani.net...
> On 2010-10-03 22:21, JF Mezei wrote:
>> Why run a browser on VMS ?
>>
>> When you need to download stuff to your VMS system, having a browser on
>> it is convenient.
>>
>
> Much easier to use some tool like HTTP_FETCH or similar then to
> run a full browser. And it's just as easy to keep SW-kits on my
> laptop ready to be used at any particular site I'm working on.
> Much faster to make a simple FTP copy then trying to get a
> WEB browser running (on VMS).
>
>> When you want to consult documentation on-line, having a browser is
>> convenient so you can cut/paste ode examples from the browser into TPU.
>
> I have absolutely no problem att all cut-n-pasting between by Windows
> browser and my Reflection session ! That was realy dumb...
>
> There is of course absolutely no *professional* reason at all to
> fight the war against browsers under VMS and there are a heap
> of reasons to run it on (some of the) most used desktop systems.
>
> Now, if you add in historical, hobbyist or similar factors into
> the pot, I don't know. But what has that to do with anything real ?
>
> Trying to use your VMS systems for WEB access *professionally*
> is just, well, unprofessional.
>
> This is one of the things I don't know why HP are putting effort
> into, WEB browsers for VMS. Totaly wasted money...

Not to the filth at HP/VMS who are lining their pockets with it! It seems 
you just can't have too many browsers for server-centric VMS according to 
them. Not 1, not 2, not th3, but 4, yes *FOUR* web-browsers for VMS! See: -
http://h71000.www7.hp.com/ebusiness/technology.html

Yet still no IPsec and all the Axis2/C money disappearing down some gSOAP 
toilet. I just can't wait for Safari/VMS and Opera/VMS. With the northern 
winter approaching I'm sure the development teams and managers could be 
looking forward to adding to last year's ski-chalet investments. Then 
there's that top-secret vPhone project! with VMSlite being installed on 
devices as small as a large shoe-box. (Cooling and sound-dampening modules 
sold seperately)

So what version of FireFox are they running on VMS now anyway? 3.6 perhaps? 
How did they get on with the Out-Of-Process-Pluggin (OOPP) code? This should 
be right up VMS's street with the sand-boxed JAVA and Flash plugins! (Oops! 
What version of JAVA are we running on VMS again?)

And how's HTML5 holding up on VMS? Anyone running some decent SVG Canvass 
graphics yet? VMS engineering implemented that abortion of a WebSocket 
interface yet?

Regards Richard Maher

PS. Shame IE[8/9] doesn't appear to have followed the Chrome/FireFox OOPP 
model (yet). I do hope there's at least a better TabProcGrowth option other 
than direct Registry edit! 

> BREAKING NEWS:
> 
> smithfarm has been discovered to be VAXman's long lost twin brother !

Thanks for the compliment. (It was a compliment, right?)

This whole troll-fest --er, I meant to say "discussion" -- got me 
thinking about one thing: the mindset that says "VMS is not good for 
<this or that>". On the one hand, it is true that VMS's engineering 
resources are limited by factors such as the dwindling number of 
qualified engineers and the number of hours they spend prattering --er, 
I meant to say "recovering" -- on c.o.v. so the question of where to 
allocate those resources is a valid one.

On the other hand, if VMS is ever to gain any traction as a viable "real 
world" alternative to Windows and Linux, there is a certain minimal (and 
shifting) feature set that it must support. What kinds of things that 
would include is debatable, but certainly a web browser (and, yes, 
something that doesn't choke on quoted-pukeable too) would be high up on 
my list.

Having to run and maintain Linux and Windows boxen alongside VMS systems 
just because VMS lacks "this or that" is sub-optimal. VMS is a 
generalized operating system. Anything that can be done on Windows or 
Linux *can* be done on VMS. Question is, will it be done. And that's a 
resource-allocation question.

On 3 okt, 11:49, "Richard Maher" <maher...@hotspamnotmail.com> wrote:
> Hi Philip,
>
> Ok, I tried not to bite but couldn't help it; why would anyone voluntarily
> be using a VMS-based solution to browse the web?
>
> Absolute-fucking-disbelief!
>
> Cheers Richard Maher

Ahem, I've got to admit that I do the same occasionally.
It is really amazing how many 'error' messages pop up, even while
browsing trusted sites. Most error messages warn you that pieces of
(Intel) code don't seem to run and hence the site thinks your computer
is broken.
Hans

smithfarm wrote:

> On the other hand, if VMS is ever to gain any traction as a viable "real 
> world" alternative to Windows and Linux, there is a certain minimal (and 
> shifting) feature set that it must support.

Rewind 10-12 years: Enters Steve Jobs to rescue Apple. He job was to
revamp the Macintosh product line and bring it back to life.  He used
"vision" to convince the board to spend the big bucks to redevelop the OS.

Bringing VMS back to life as a server OS would take less effort than it
took Apple to resurrect itself. Port it to the 64 bit 8086 and bring in
modern stuff like IPsec and true IPv6 support.

Porting the browser is in fact fairly smart because it also forced the
porting of much middleware such as php, perl and a buch of others that
are used to build the Mozilla product. But it must remain a continuous
effort to ensure that VMS has up-to-date middleware.

Also, if you need another OS to manage a server OS, you might as well
use that other OS as your server OS.


At this point in time we have to be realistic as painful as it may be.
HP has had 10 years to change its mind about VMS.  It hasn't. It won't.

With VMS management and engineering gone, there is nobody to fight for
VMS at HP.  And when the new guy inquires about HP's products, he will
probably be told that it is too late to revive VMS and that the
remaining VMS customers are happy with VMS being on maintenance mode
with limited new development. There will be nobody to counter this.

Existing mamagement like Livermore will arrange for the new guy to meet
customers whose opinion of VMS matches what HP's policy on VMS is.

I don't know why anyone would still have hopes that HP would start a
major turn around for VMS.

In article <i8bip8$tnc$1@speranza.aioe.org>, "Richard Maher"
<maher_rj@hotspamnotmail.com> writes:
> 
> Not to the filth at HP/VMS who are lining their pockets with it! It seems 
> you just can't have too many browsers for server-centric VMS

that's your very narrow minded personal view.

> according to 
> them. Not 1, not 2, not th3, but 4, yes *FOUR* web-browsers for VMS! See: -
> http://h71000.www7.hp.com/ebusiness/technology.html

Somehow I've the impression you're somewhat read impaired.
If you'd actually bothered to lookup that link you would have
noticed that these are not four current browsers developed in parallel.
One is an ancient (frozen) VAX one, a second one (Mozilla based)
is redirected to the third one (Seamonkey based),
which leaves only the fourth one (Firefox for Itanic only)
as an extra service to bitch about.
I wonder how often I will have to bring that to your attentiom.

Moreover, as others have pointed out repeatedly,
porting a browser requires a gazillion of support libraries
and infrastructure
which one needs anyway if one wants to stay in the game
with current open source software. So this effort
is by no means wasted.

In article <4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com>, JF Mezei
<jfmezei.spamnot@vaxination.ca> writes:
> 
> Also, if you need another OS to manage a server OS, you might as well
> use that other OS as your server OS.
> 

exactly.
Moreover, to come up with another of those popular
car analogies: what some people suggest here is
to sell a car restricted to engine and tank ("server-centric").
If you want to actually drive it,
you'd have to buy/use a second one,
which will give you seats, tires and a driving wheel,
so you can hook up the first one.
I think most, if not all customers would just
buy the second, complete one, even if it 
has a "weaker" (by whatever measure) engine
than the first one.

On 10/4/2010 2:15 AM, smithfarm wrote:

> Having to run and maintain Linux and Windows boxen alongside VMS systems
> just because VMS lacks "this or that" is sub-optimal. VMS is a
> generalized operating system. Anything that can be done on Windows or
> Linux *can* be done on VMS. Question is, will it be done. And that's a
> resource-allocation question.

I wouldn't hold my breath!!

Competing O/S run on cheap commodity hardware.  They're NOT VMS of 
course but they generally get the job done.

Last time I looked, a commercial single user VMS license cost ~2K.  It's 
been a LONG time since I looked at the price of a FORTRAN compiler for 
VMS but when I did, it was something like $2K per seat.

Shall we look at what that $4K will buy in the X86 world?  That's right! 
  It buys the computer, the O/S, a word processor, a spreadsheet,
a database. . . .

In article <xrGdnah-dpDgQzTRnZ2dnUVZ_uudnZ2d@giganews.com>, "Richard B. Gilbert"
<rgilbert88@comcast.net> writes:
> 
> I wouldn't hold my breath!!
> 
> Competing O/S run on cheap commodity hardware.  They're NOT VMS of 
> course but they generally get the job done.
> 
> Last time I looked, a commercial single user VMS license cost ~2K.  It's 
> been a LONG time since I looked at the price of a FORTRAN compiler for 
> VMS but when I did, it was something like $2K per seat.
> 
> Shall we look at what that $4K will buy in the X86 world?  That's right! 
>   It buys the computer, the O/S, a word processor, a spreadsheet,
> a database. . . .

It's not about competing with the M$ world pricewise on the retail market.
It's about offering a complete OS which allows the user and/or admin
to perform 90+x% of everyday tasks (which IMHO would include
emailing and web-browsing) without the necessity to resort
to a second platform. One probably doesn't need 100% feature parity
with youtube-capable browsers, but mileages may vary, of course.

BTW, given that VMS as well as Itanic are practically extinct in technical
computing, a Fortran compiler is probably even less urgent than a web browser.
And there's always gcc.

In article <i8cphe$qr1$1@lnx107.hrz.tu-darmstadt.de>,
	m.kraemer@gsi.de (Michael Kraemer) writes:
> In article <xrGdnah-dpDgQzTRnZ2dnUVZ_uudnZ2d@giganews.com>, "Richard B. Gilbert"
> <rgilbert88@comcast.net> writes:
>> 
>> I wouldn't hold my breath!!
>> 
>> Competing O/S run on cheap commodity hardware.  They're NOT VMS of 
>> course but they generally get the job done.
>> 
>> Last time I looked, a commercial single user VMS license cost ~2K.  It's 
>> been a LONG time since I looked at the price of a FORTRAN compiler for 
>> VMS but when I did, it was something like $2K per seat.
>> 
>> Shall we look at what that $4K will buy in the X86 world?  That's right! 
>>   It buys the computer, the O/S, a word processor, a spreadsheet,
>> a database. . . .
> 
> It's not about competing with the M$ world pricewise on the retail market.

Well, if we ignore the M$ world that $4k buys a hell of a lot more.
Including all the above mentioned pieces.  Probably a dozen systems.
Or a couple of killer servers.

> It's about offering a complete OS which allows the user and/or admin
> to perform 90+x% of everyday tasks (which IMHO would include
> emailing and web-browsing) without the necessity to resort
> to a second platform. One probably doesn't need 100% feature parity
> with youtube-capable browsers, but mileages may vary, of course.

But one can have all of that at a fraction of the cost of VMS.  But then,
you knew that and just choose to turn a deaf ear.

> 
> BTW, given that VMS as well as Itanic are practically extinct in technical
> computing, a Fortran compiler is probably even less urgent than a web browser.
> And there's always gcc.

Actually, I see jobs looking for Fortran skills all the time.  Especially
in government jobs at palces like NOAA and DOD and in contractors as well.
I guess Fortran is as dead as COBOL.  :-)  But to get back to the point,
both Fortran and COBOL as well as a dozen other languages come free in
the world outside of VMS.  Doesn't mean VMS isn't better, just not in any
way competitive.

bill


-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

> BTW, given that VMS as well as Itanic are practically extinct in technical
> computing, a Fortran compiler is probably even less urgent than a web browser.
> And there's always gcc.

Yes, I forgot to mention in my earlier post: based on my futile search 
for used I64 hardware from a reputable local dealer, it does appear that 
a port of OpenVMS to amd64 is the only way forward. (Barring the 
emergence of a better platform, of course.) There's no way an OS can 
survive, let alone thrive, if there's no affordable hardware to run it on.

In article <8gu9naF9f0U3@mid.individual.net>, billg999@cs.uofs.edu (Bill
Gunshannon) writes:
> 
> But one can have all of that at a fraction of the cost of VMS.  But then,
> you knew that and just choose to turn a deaf ear.

I'm not advocating VMS, of course, just that it would be
more attractive if it came as a complete OS standing on its own feet. 
 
> > 
> > BTW, given that VMS as well as Itanic are practically extinct in technical
> > computing, a Fortran compiler is probably even less urgent than a web browser.
> > And there's always gcc.
> 
> Actually, I see jobs looking for Fortran skills all the time.  Especially
> in government jobs at palces like NOAA and DOD and in contractors as well.
> I guess Fortran is as dead as COBOL.  :-) 

I didn't say Fortran is dead. 
VMS and Itanic are, at least in technical
computing, which makes it less urgent
to offer a native Fortran/Itanic compiler for VMS. 

"JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message 
news:4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com...
> smithfarm wrote:

>
> Bringing VMS back to life as a server OS would take less effort than it
> took Apple to resurrect itself. Port it to the 64 bit 8086 and bring in
> modern stuff like IPsec and true IPv6 support.
>

I've said this 10000 times, and I'll repeat it 10000 more.  Porting VMS to 
x86-64 for any and all of it's benefits - will make NO DIFFERENCE.  Or 
rather, the difference would be so marginal that it would have no real 
meaning.

Nobody really cares what OS is under the hood on Apple.  What they care 
about are the applications that run on it.  They also remain a tiny niche of 
the desktop market.  Nor, may I say are there computers what is leading 
Apple's business anymore.  iPhones and iPods are the lifeblood of Apple.

> I've said this 10000 times, and I'll repeat it 10000 more.  Porting VMS to 
> x86-64 for any and all of it's benefits - will make NO DIFFERENCE.  Or 
> rather, the difference would be so marginal that it would have no real 
> meaning.
> 
> Nobody really cares what OS is under the hood on Apple.  What they care 
> about are the applications that run on it.  They also remain a tiny niche of 
> the desktop market.  Nor, may I say are there computers what is leading 
> Apple's business anymore.  iPhones and iPods are the lifeblood of Apple.

I'm the last one who would argue with you, Fred.

This begs a hypothetical question (I apologize in advance if it's been 
posed here a million times before): VMS is touted for its stability and 
security. If, in some parallel universe, VMS were the dominant OS 
running on 90% of end-user machines, would the security situation really 
be any better? Would there be fewer (and smaller) botnets, etc., less 
port scanning, less spam, etc.?

"smithfarm" <presnypreklad@gmail.com> wrote in message 
news:i8d27r$2gh$1@speranza.aioe.org...
>> I've said this 10000 times, and I'll repeat it 10000 more.  Porting VMS 
>> to x86-64 for any and all of it's benefits - will make NO DIFFERENCE.  Or 
>> rather, the difference would be so marginal that it would have no real 
>> meaning.
>>
>> Nobody really cares what OS is under the hood on Apple.  What they care 
>> about are the applications that run on it.  They also remain a tiny niche 
>> of the desktop market.  Nor, may I say are there computers what is 
>> leading Apple's business anymore.  iPhones and iPods are the lifeblood of 
>> Apple.
>
> I'm the last one who would argue with you, Fred.
>
> This begs a hypothetical question (I apologize in advance if it's been 
> posed here a million times before): VMS is touted for its stability and 
> security. If, in some parallel universe, VMS were the dominant OS running 
> on 90% of end-user machines, would the security situation really be any 
> better? Would there be fewer (and smaller) botnets, etc., less port 
> scanning, less spam, etc.?

IMHO - security got out-of-control when it became possible to send methods 
as opposed to data for useful purposes.  When it became possible for a 
email, or a web site to execute code without the explicit control of the 
user - the gates were wide open.

The PC for most people is a tool they use, and that they don't understand - 
which is as it should be - you should not need to know how the engine in 
your car works to drive one.  But that also means security settings will be 
beyond most users to understand or control.  If a user gets a pop up that 
says "Application foo is trying to read bar, do you want to allow it" - they 
have no idea what to do.  What they want is to do all these useful and cool 
things, which as a side effect introduces high risks.

From an OS perspective VMS is safer than UNIX-like OSes for many aspects of 
operation because of roots in it's design - part of which was language and 
calling conventions.  VMS would never have chosen "malloc" as a memory 
allocation API.  Or depended on zero termination for strings.  The fact that 
we now have a lot of C code makes us vulnerable - but the good news is that 
the OS calling conventions haven't changed - so the effect tends to be more 
around application crashers.  There is little we can do about TCPIP - it 
won, OSI lost.  We have to live and try to deal with the problems that come 
along with it.  If VMS had won, TCPIP would not have.

Apple with a UNIX-like OS under the hood has fewer issues than windows.  VMS 
has less than UNIX-like OS's.  That isn't to say that we don't have security 
bugs.  I've fixed at least one within the last decade.  But they are 
generally less likely - and there are fewer eyes and hands trying to find 
them and exploit them.  If VMS had beaten UNIX - then more would be found. 
But the question is if VMS had beaten UNIX - would C/C++ languages be the 
dominant languages?  Or Pascal and PL/I (Fortran, Bliss, Ada, ...)?

In article <4ca8e5b3$0$17470$c3e8da3$b280bf18@news.astraweb.com>, JF Mezei <jfmezei.spamnot@vaxination.ca> writes:
> Why run a browser on VMS ?
> 
> When you need to download stuff to your VMS system, having a browser on
> it is convenient.
> 
> When you want to consult documentation on-line, having a browser is
> convenient so you can cut/paste ode examples from the browser into TPU.

   When you don't want to worry about what viri et. al. you might
   encounter, having a browser on VMS is great.

In article <i87qoo$2ep$1@online.de>, helbig@astro.multiCLOTHESvax.de (Phillip Helbig---undress to reply) writes:
> I've been running Mozilla (SWB) V1.7-11 which I downloaded from HP for a
> while.  It's been sufficient for my needs and even on quite old DEC
> (yes, DEC, not even Compaq) hardware (EV56) it is fast enough for my 
> needs.
> 
> A site I sometimes visit now crashes it.  Yes, there were changes to the 
> site and yes, the programming of the page in question is probably not 
> the best, but still, an application shouldn't crash even if fed invalid 
> input.  The error is below.

   I don't know if it is the same crash, but I had a similar problem
   when I left in the no-longer-working FLASH extension for Mozilla
   on VMS.  Guess why it's supposed to be removed?

On 10/4/2010 10:49 AM, Michael Kraemer wrote:
> In article<xrGdnah-dpDgQzTRnZ2dnUVZ_uudnZ2d@giganews.com>, "Richard B. Gilbert"
> <rgilbert88@comcast.net>  writes:
>>
>> I wouldn't hold my breath!!
>>
>> Competing O/S run on cheap commodity hardware.  They're NOT VMS of
>> course but they generally get the job done.
>>
>> Last time I looked, a commercial single user VMS license cost ~2K.  It's
>> been a LONG time since I looked at the price of a FORTRAN compiler for
>> VMS but when I did, it was something like $2K per seat.
>>
>> Shall we look at what that $4K will buy in the X86 world?  That's right!
>>    It buys the computer, the O/S, a word processor, a spreadsheet,
>> a database. . . .
>
> It's not about competing with the M$ world pricewise on the retail market.
> It's about offering a complete OS which allows the user and/or admin
> to perform 90+x% of everyday tasks (which IMHO would include
> emailing and web-browsing) without the necessity to resort
> to a second platform. One probably doesn't need 100% feature parity
> with youtube-capable browsers, but mileages may vary, of course.
>
> BTW, given that VMS as well as Itanic are practically extinct in technical
> computing, a Fortran compiler is probably even less urgent than a web browser.
> And there's always gcc.

If  a web browser is so urgent, why has no one ported Mozilla Firefox to 
VMS?  And how about the Thunderbird mail/news client?  Could it be that 
nobody cares any longer?

I don't miss having a browser on VMS, I can always use the one on my PC.
Ditto for mail/news and the Thunderbird mail client.

"Richard B. Gilbert" <rgilbert88@comcast.net> wrote in message 
news:yKKdnRFCD7DMtDfRnZ2dnUVZ_sOdnZ2d@giganews.com...
> On 10/4/2010 10:49 AM, Michael Kraemer wrote:

>
> If  a web browser is so urgent, why has no one ported Mozilla Firefox to 
> VMS?  And how about the Thunderbird mail/news client?  Could it be that 
> nobody cares any longer?
>

Seamonkey is roughly the combination of Firefox and Thunderbird.  So its 
less resources to do just one of these than two or three of them.

In article <i8d3ae$1ak$1@usenet01.boi.hp.com>,
	"FredK" <fred.nospam@dec.com> writes:
> 
> From an OS perspective VMS is safer than UNIX-like OSes for many aspects of 
> operation because of roots in it's design - part of which was language and 
> calling conventions.  VMS would never have chosen "malloc" as a memory 
> allocation API.  Or depended on zero termination for strings. 

When are we going to put this myth to sleep.  Unix didn't invent null
terminated strings.  DEC was perfectly happy with them as well.  All
of the non-Unix PDP-11 OSes had them and no matter how hard people here
try to deny it, so did VMS, from the very beginning (and, I'll bet it's
still there in MACRO for anyone who wants to use it!)

-------------------------------------------------------------------------------
ASCIZ

Zero-terminated ASCII string storage directive 

--------------------------------------------------------------------------------

Format
..ASCIZ string 


--------------------------------------------------------------------------------

Parameter
string
A delimited ASCII string. 

--------------------------------------------------------------------------------

Description
..ASCIZ performs the same function as .ASCII, except that .ASCIZ appends a null byte as the final character of the string. When a list or text string is created with an .ASCIZ directive, you need only perform a search for the null character in the last byte to determine the end of the string. 

--------------------------------------------------------------------------------

Example
 

 FF=12                                  ; Define FF

        .ASCIZ  /ABCDEF/                ; 6 characters in string,
                                        ;   7 bytes of data
        .ASCIZ  /A/<FF>/B/              ; 3 characters in strings

And, a famous example of its use!!  :-)

------------------------------------------------------------------------------
       
; Vax VMS version of 99 Bottles of Beer
       .TITLE BOTTLES
BOTT:  .ASCIZ / BOTTLES OF BEER ON THE WALL,/
BOTT2: .ASCIZ / BOTTLES OF BEER, ONE DOWN, PASS IT AROUND/
BOTT3: .ASCIZ / BOTTLES OF BEER ON THE WALL./
ZERO:  .LONG 48

       .ENTRY START,0
       MOVL #^A'9',R2
       MOVL #^A'9',R3
LOOP:  .TTYOUT R3
       .TTYOUT R2
       .PUTSTR BOTT
       .TTYOUT R3
       .TTYOUT R2
       .PUTSTR BOTT2
       CMPL R2,ZERO
       BEQL DECR
CONT1: DECL R2
       .TTYOUT R3
       .TTYOUT R2
       .PUTSTR BOTT3
       CMPL R3,ZERO
       BEQL CHECK
       BRB LOOP
DECR:  DECL R3
       MOVL #^A':',R2
       BRB CONT1
CHECK: CMPL R2,ZERO
       BGTR LOOP
       $EXIT_S
       .END START

------------------------------------------------------------------------------

Anybody want to put any money on wether or not .PUTSTR does any checking
other than looking for a terminating 0 byte?

Yes, I know that there are other ways to handle strings in VMS.  There
are in Unix, too.  Pretty much an infinte number of ways limited only
by the programmers imagination.

bill


-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

"Bill Gunshannon" <billg999@cs.uofs.edu> wrote in message 
news:8guoccF6t8U1@mid.individual.net...
> In article <i8d3ae$1ak$1@usenet01.boi.hp.com>,
> "FredK" <fred.nospam@dec.com> writes:
>>
>> From an OS perspective VMS is safer than UNIX-like OSes for many aspects 
>> of
>> operation because of roots in it's design - part of which was language 
>> and
>> calling conventions.  VMS would never have chosen "malloc" as a memory
>> allocation API.  Or depended on zero termination for strings.
>
> When are we going to put this myth to sleep.  Unix didn't invent null
> terminated strings.  DEC was perfectly happy with them as well.  All
> of the non-Unix PDP-11 OSes had them and no matter how hard people here
> try to deny it, so did VMS, from the very beginning (and, I'll bet it's
> still there in MACRO for anyone who wants to use it!)
>

Bill, I didn't mean to imply UNIX or C "invented" zwro terminated strings. 
C made it the language standard, UNIX enshrined it in it's interfaces. 
Despite .ASCIZ - find it used as part of the calling standard.  It was 
stupidity on a grand scale. 

FredK wrote:

> I've said this 10000 times, and I'll repeat it 10000 more.  Porting VMS to 
> x86-64 for any and all of it's benefits - will make NO DIFFERENCE.  Or 
> rather, the difference would be so marginal that it would have no real 
> meaning.

So Mr FredK, as one of the few remaining real VMS engineers left at HP,
if the new CEO were to ask your opinion, this is what you would tell him?

You would rather see VMS restricted to an unwanted IA64 platform whose
future is uncertain and performance now below par compared to the 8086 ?

FredK wrote:

> calling conventions.  VMS would never have chosen "malloc" as a memory 
> allocation API. 

I ask this naively. What is wrong with malloc ?  You give it a size, it
returns a pointer to a memory block allocated to you of that size.


> Apple with a UNIX-like OS under the hood has fewer issues than windows.  VMS 
> has less than UNIX-like OS's.  That isn't to say that we don't have security 
> bugs.

One metric used these days is the number of days between the discovery
of a bug and a patch being made available. (aka: length of vulnerability).

For that metric, VMS is terrible. Thankfully, VMS is so unimportant that
its security bugs don't get reported to CERT or other widely publiscised
mediums.

> But the question is if VMS had beaten UNIX - would C/C++ languages be the 
> dominant languages?  Or Pascal and PL/I (Fortran, Bliss, Ada, ...)?

The world would be all in BLISS and DIBOL :-) :-) :-)

"JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message 
news:4caa383d$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
> FredK wrote:
>
>> I've said this 10000 times, and I'll repeat it 10000 more.  Porting VMS 
>> to
>> x86-64 for any and all of it's benefits - will make NO DIFFERENCE.  Or
>> rather, the difference would be so marginal that it would have no real
>> meaning.
>
> So Mr FredK, as one of the few remaining real VMS engineers left at HP,
> if the new CEO were to ask your opinion, this is what you would tell him?
>
> You would rather see VMS restricted to an unwanted IA64 platform whose
> future is uncertain and performance now below par compared to the 8086 ?

I would simply ask to be treated as a business.  If we meet our goals and 
are profitable, don't reward us by reducing OpEx (operating expenses).  If 
we are profitable, let us invest a percentage of that profit in growth.

That is all I could ask of any CEO.

The platform VMS runs on is not really material.  If at some mythical point 
in time Itanium stops being produced, then it has meaning (not just to VMS). 
I personally also believe that a small, cheap developers machine is 
important.  But I don't believe that if we made a $1000 desktop VMS system, 
that it would change the market dynamics.

Don't get me wrong in thinking that I would not like to see a x86-64 VMS 
box - I just do not believe it is the magic bullet you apparently think it 
is.

On Oct 4, 8:31=A0pm, billg...@cs.uofs.edu (Bill Gunshannon) wrote:
> In article <i8d3ae$1a...@usenet01.boi.hp.com>,
> =A0 =A0 =A0 =A0 "FredK" <fred.nos...@dec.com> writes:
>
>
>
> > From an OS perspective VMS is safer than UNIX-like OSes for many aspect=
s of
> > operation because of roots in it's design - part of which was language =
and
> > calling conventions. =A0VMS would never have chosen "malloc" as a memor=
y
> > allocation API. =A0Or depended on zero termination for strings.
>
> When are we going to put this myth to sleep. =A0Unix didn't invent null
> terminated strings. =A0DEC was perfectly happy with them as well. =A0All
> of the non-Unix PDP-11 OSes had them and no matter how hard people here
> try to deny it, so did VMS, from the very beginning (and, I'll bet it's
> still there in MACRO for anyone who wants to use it!)
>
> -------------------------------------------------------------------------=
------
> ASCIZ
>
> Zero-terminated ASCII string storage directive
>
> -------------------------------------------------------------------------=
-------
>
> Format
> .ASCIZ string
>
> -------------------------------------------------------------------------=
-------
>
> Parameter
> string
> A delimited ASCII string.
>
> -------------------------------------------------------------------------=
-------
>
> Description
> .ASCIZ performs the same function as .ASCII, except that .ASCIZ appends a=
 null byte as the final character of the string. When a list or text string=
 is created with an .ASCIZ directive, you need only perform a search for th=
e null character in the last byte to determine the end of the string.
>
> -------------------------------------------------------------------------=
-------
>
> Example
>
> =A0FF=3D12 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0 =A0; Define FF
>
> =A0 =A0 =A0 =A0 .ASCIZ =A0/ABCDEF/ =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0; 6 cha=
racters in string,
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0 =A0 ; =A0 7 bytes of data
> =A0 =A0 =A0 =A0 .ASCIZ =A0/A/<FF>/B/ =A0 =A0 =A0 =A0 =A0 =A0 =A0; 3 chara=
cters in strings
>
> And, a famous example of its use!! =A0:-)
>
> -------------------------------------------------------------------------=
-----
>
> ; Vax VMS version of 99 Bottles of Beer
> =A0 =A0 =A0 =A0.TITLE BOTTLES
> BOTT: =A0.ASCIZ / BOTTLES OF BEER ON THE WALL,/
> BOTT2: .ASCIZ / BOTTLES OF BEER, ONE DOWN, PASS IT AROUND/
> BOTT3: .ASCIZ / BOTTLES OF BEER ON THE WALL./
> ZERO: =A0.LONG 48
>
> =A0 =A0 =A0 =A0.ENTRY START,0
> =A0 =A0 =A0 =A0MOVL #^A'9',R2
> =A0 =A0 =A0 =A0MOVL #^A'9',R3
> LOOP: =A0.TTYOUT R3
> =A0 =A0 =A0 =A0.TTYOUT R2
> =A0 =A0 =A0 =A0.PUTSTR BOTT
> =A0 =A0 =A0 =A0.TTYOUT R3
> =A0 =A0 =A0 =A0.TTYOUT R2
> =A0 =A0 =A0 =A0.PUTSTR BOTT2
> =A0 =A0 =A0 =A0CMPL R2,ZERO
> =A0 =A0 =A0 =A0BEQL DECR
> CONT1: DECL R2
> =A0 =A0 =A0 =A0.TTYOUT R3
> =A0 =A0 =A0 =A0.TTYOUT R2
> =A0 =A0 =A0 =A0.PUTSTR BOTT3
> =A0 =A0 =A0 =A0CMPL R3,ZERO
> =A0 =A0 =A0 =A0BEQL CHECK
> =A0 =A0 =A0 =A0BRB LOOP
> DECR: =A0DECL R3
> =A0 =A0 =A0 =A0MOVL #^A':',R2
> =A0 =A0 =A0 =A0BRB CONT1
> CHECK: CMPL R2,ZERO
> =A0 =A0 =A0 =A0BGTR LOOP
> =A0 =A0 =A0 =A0$EXIT_S
> =A0 =A0 =A0 =A0.END START
>
> -------------------------------------------------------------------------=
-----
>
> Anybody want to put any money on wether or not .PUTSTR does any checking
> other than looking for a terminating 0 byte?
>
> Yes, I know that there are other ways to handle strings in VMS. =A0There
> are in Unix, too. =A0Pretty much an infinte number of ways limited only
> by the programmers imagination.
>
> bill
>
> --
> Bill Gunshannon =A0 =A0 =A0 =A0 =A0| =A0de-moc-ra-cy (di mok' ra see) n. =
=A0Three wolves
> billg...@cs.scranton.edu | =A0and a sheep voting on what's for dinner.
> University of Scranton =A0 |
> Scranton, Pennsylvania =A0 | =A0 =A0 =A0 =A0 #include <std.disclaimer.h> =
=A0

Of course it'd be silly to deny that null strings were available on
VMS. Some people even used them.

It's even sillier to claim that null terminated strings were ever the
natural or default choice for VMS languages or compilers. Descriptors
were the natural default choice. Any mainstream VMS language (ie
anything other than C) made using descriptors trivially simple
(transparent, even).

To go with that, the VMS RTL and LIB$whatever and friends made using
descriptors trivially easy too.

Why would any sensible person use anything other than descriptors?

Well the usual reason in a non-trivial example would be because they
needed to interface to routines requiring K+R-style (UNIX-style, if
you wish) string handling routines.

Null terminated strings in anything but trivial code are at best a
tedious opprtunity for error, at worst a nightmare. With the benefit
of hindsight, K+R got this one wrong.

Properly designed VMS-native code has little or no need for null
terminated strings, and you can get all kinds of idiot-resistance done
for you by the compiler, the RTL, or both.

Which UNIX man pages (or pages from K+R, choose your version) document
the other methods of string handling to which you refer?

So yes, let's put it to sleep.

"Anybody want to put any money on wether or not .PUTSTR does any
checking other than looking for a terminating 0 byte?"

What other checking would you want it to do, given that some
foreigners have the audacity to want to use all eight bits in a byte
these days?

"JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message 
news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
> FredK wrote:
>
>> calling conventions.  VMS would never have chosen "malloc" as a memory
>> allocation API.
>
> I ask this naively. What is wrong with malloc ?  You give it a size, it
> returns a pointer to a memory block allocated to you of that size.
>

It also doesn't return an error code.

>
>> Apple with a UNIX-like OS under the hood has fewer issues than windows. 
>> VMS
>> has less than UNIX-like OS's.  That isn't to say that we don't have 
>> security
>> bugs.
>
> One metric used these days is the number of days between the discovery
> of a bug and a patch being made available. (aka: length of vulnerability).
>
> For that metric, VMS is terrible. Thankfully, VMS is so unimportant that
> its security bugs don't get reported to CERT or other widely publiscised
> mediums.
>

That VMS isn't responsive to CERT?  Or that we don't fix reported bugs? 
Many of which are by implication and not observation?

In article <i8asds$f1b$1@news.albasani.net>, Jan-Erik Soderholm
<jan-erik.soderholm@telia.com> writes: 

> On 2010-10-03 22:21, JF Mezei wrote:
> > Why run a browser on VMS ?
> >
> > When you need to download stuff to your VMS system, having a browser on
> > it is convenient.
> 
> Much easier to use some tool like HTTP_FETCH or similar then to
> run a full browser. 

If one knows what one wants, sure, WGET, CURL, whatever.  However, take 
as an example downloading SeaMonkey from the HP website: browse here, 
fill in a form there, read some background stuff, download somewhere 
else etc.

> And it's just as easy to keep SW-kits on my
> laptop ready to be used at any particular site I'm working on.

That's a different situation than doing everything on one VMS cluster.

> Much faster to make a simple FTP copy then trying to get a
> WEB browser running (on VMS).

Is it difficult?  When I last installed Mozilla on VMS, it took just a 
few minutes.

> Now, if you add in historical, hobbyist or similar factors into
> the pot, I don't know. But what has that to do with anything real ?
> 
> Trying to use your VMS systems for WEB access *professionally*
> is just, well, unprofessional.

Depends on the profession.

In article <i8asge$jl6$1@speranza.aioe.org>, smithfarm
<presnypreklad@gmail.com> writes: 

> > For those who want to go (back) to VMS mail, this blast from the past 
> > provides a way!  (I hope Elliott still follows the group.  There might 
> > be some undocumented TECO code below which will start up a lawnmower.)
> > 
> Does it handle quoted-pukeable? Binary attachments?

I've never used it myself; I started using VMS MAIL about the same time 
I got internet access.

In article <i8cphe$qr1$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de
(Michael Kraemer) writes: 

> It's not about competing with the M$ world pricewise on the retail market.
> It's about offering a complete OS which allows the user and/or admin
> to perform 90+x% of everyday tasks (which IMHO would include
> emailing and web-browsing) without the necessity to resort
> to a second platform. One probably doesn't need 100% feature parity
> with youtube-capable browsers, but mileages may vary, of course.

Look up the Dilbert cartoon with the punch line "It was a funny one 
about a cat."

On Mon, 04 Oct 2010 17:01:02 -0400, FredK wrote:

> "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message
> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>> FredK wrote:
>>
>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>> allocation API.
>>
>> I ask this naively. What is wrong with malloc ?  You give it a size, it
>> returns a pointer to a memory block allocated to you of that size.
>>
>>
> It also doesn't return an error code.

If it complies with the C standard, it does.

-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

FredK wrote:

> Don't get me wrong in thinking that I would not like to see a x86-64 VMS 
> box - I just do not believe it is the magic bullet you apparently think it 
> is.

Porting VMS to the 8086 is not a magic bullet. Changing HP's mind on VMS
is the magic bullet.

Porting to the 8086 would be a sign that HP is changing its mind about
VMS and would likely be accompanied with other steps such as allowing
marketing of VMS etc.

When the "HP is killing VMS" noise gets too loud, HP will simply take a
token action (such as one time marketing) to quiet that noise. This has
happened since the palmer era. So token actions are no longer sufficient
to convince people of a real change.

The port of VMS to 8086 would be a substantial sign of a commitment one
that would not be short lived.

"Bob Eager" <rde42@spamcop.net> wrote in message 
news:8gv1rnFefcU5@mid.individual.net...
> On Mon, 04 Oct 2010 17:01:02 -0400, FredK wrote:
>
>> "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message
>> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>>> FredK wrote:
>>>
>>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>>> allocation API.
>>>
>>> I ask this naively. What is wrong with malloc ?  You give it a size, it
>>> returns a pointer to a memory block allocated to you of that size.
>>>
>>>
>> It also doesn't return an error code.
>
> If it complies with the C standard, it does.
>

foo = malloc(size);

The only "error" return is NULL.   Returning NULL provides me with zero 
information that can be either used to correct the problem, or instruct the 
user how to correct the problem - other than guessing.  So to be more long 
winded, if there can be more than one possible error failure, a function 
should return a seperate error code.

malloc was simply an example of the generic aversion to error handling, 
followed by kludges to provide errors.

JF Mezei <jfmezei.spamnot@vaxination.ca> wrote:
 
> Porting VMS to the 8086 is not a magic bullet. Changing HP's mind on VMS
> is the magic bullet.
 
> Porting to the 8086 would be a sign that HP is changing its mind about
> VMS and would likely be accompanied with other steps such as allowing
> marketing of VMS etc.

No-one is going to port to the 8086.  Most likely not even IA32.

x86-64 seems like the best hope, but probably not even that.

-- glen

On Mon, 04 Oct 2010 18:32:21 -0400, FredK wrote:

> "Bob Eager" <rde42@spamcop.net> wrote in message
> news:8gv1rnFefcU5@mid.individual.net...
>> On Mon, 04 Oct 2010 17:01:02 -0400, FredK wrote:
>>
>>> "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message
>>> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>>>> FredK wrote:
>>>>
>>>>> calling conventions.  VMS would never have chosen "malloc" as a
>>>>> memory allocation API.
>>>>
>>>> I ask this naively. What is wrong with malloc ?  You give it a size,
>>>> it returns a pointer to a memory block allocated to you of that size.
>>>>
>>>>
>>> It also doesn't return an error code.
>>
>> If it complies with the C standard, it does.
>>
>>
> foo = malloc(size);
> 
> The only "error" return is NULL.   Returning NULL provides me with zero
> information that can be either used to correct the problem, or instruct
> the user how to correct the problem - other than guessing.  So to be
> more long winded, if there can be more than one possible error failure,
> a function should return a seperate error code.
> 
> malloc was simply an example of the generic aversion to error handling,
> followed by kludges to provide errors.

Or an example of a distorted hate campaign...

If you are determined to find fault, then  of course you will do so.



-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

"Bob Eager" <rde42@spamcop.net> wrote in message 
news:8gv5okFefcU7@mid.individual.net...
> On Mon, 04 Oct 2010 18:32:21 -0400, FredK wrote:

[...]

>
> Or an example of a distorted hate campaign...
>
> If you are determined to find fault, then  of course you will do so.
>

Oh geeze.  Yeah, it's an irrational hate campaign.  What exactly is it that 
I am supposed to hate?  Bad interfaces?  Yes.  You got me.

On Mon, 04 Oct 2010 19:32:49 -0400, FredK wrote:

> "Bob Eager" <rde42@spamcop.net> wrote in message
> news:8gv5okFefcU7@mid.individual.net...
>> On Mon, 04 Oct 2010 18:32:21 -0400, FredK wrote:
> 
> [...]
> 
> 
>> Or an example of a distorted hate campaign...
>>
>> If you are determined to find fault, then  of course you will do so.
>>
>>
> Oh geeze.  Yeah, it's an irrational hate campaign.  What exactly is it
> that I am supposed to hate?  Bad interfaces?  Yes.  You got me.

Q.E.D.




-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

"Bob Eager" <rde42@spamcop.net> wrote in message 
news:8gv6qsFefcU8@mid.individual.net...
> On Mon, 04 Oct 2010 19:32:49 -0400, FredK wrote:
>
>> "Bob Eager" <rde42@spamcop.net> wrote in message
>> news:8gv5okFefcU7@mid.individual.net...
>>> On Mon, 04 Oct 2010 18:32:21 -0400, FredK wrote:
>>
>> [...]
>>
>>
>>> Or an example of a distorted hate campaign...
>>>
>>> If you are determined to find fault, then  of course you will do so.
>>>
>>>
>> Oh geeze.  Yeah, it's an irrational hate campaign.  What exactly is it
>> that I am supposed to hate?  Bad interfaces?  Yes.  You got me.
>
> Q.E.D.
>

LOL.  Thanks for your insight.  Please defend the language decision to punt 
on strings, and the subsequent brain dead library string routines.  Defend 
the random way errors are reported and how programmers consistently 
mis-program even simple things like getting a string from a file.

The string decisions themselves have cost billions of dollars in program 
errors that led to security bugs and system and application crashes.

QED 

On Oct 4, 11:13=A0pm, Bob Eager <rd...@spamcop.net> wrote:
> On Mon, 04 Oct 2010 17:01:02 -0400, FredK wrote:
> > "JF Mezei" <jfmezei.spam...@vaxination.ca> wrote in message
> >news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
> >> FredK wrote:
>
> >>> calling conventions. =A0VMS would never have chosen "malloc" as a mem=
ory
> >>> allocation API.
>
> >> I ask this naively. What is wrong with malloc ? =A0You give it a size,=
 it
> >> returns a pointer to a memory block allocated to you of that size.
>
> > It also doesn't return an error code.
>
> If it complies with the C standard, it does.
>
> --
> Use the BIG mirror service in the UK:
> =A0http://www.mirrorservice.org
>
> *lightning protection* - a w_tom conductor

Fred may not be 100% accurate in saying no error code is returned by
malloc(), but he is *right* about the Unix approach to error reporting
in general. Windows (with GetLastError()) is no better either.

Find a unix and say
man 3 malloc

Here is what it says about what happens on error (my summary, and
consistent with various formal UNIX specs)

malloc shall return NULL on error
malloc shall set errno to ENOMEM on error

So anyone arguing a *useful* error code is returned is onto a loser.
Such is UNIX. RT11 and RSX had their equivalents of errno (the JSW and
the DSW respectively) but the user experience was rather more
friendly.

Unix has no "Error Messages and Recovery Procedures" manual, and for
most applications errno.h might as well not exist either. How many
different UNIX error messages from different applications can you get
for the one simple error where RT11, RSX, and VMS would all
consistently have said something clear and simple like "file not
found" (or local language equivalent if applicable)?

John Wallace <johnwallace4@yahoo.co.uk> wrote:
(snip on malloc error messages)
 
> Fred may not be 100% accurate in saying no error code is returned by
> malloc(), but he is *right* about the Unix approach to error reporting
> in general. Windows (with GetLastError()) is no better either.
> 
> Here is what it says about what happens on error (my summary, and
> consistent with various formal UNIX specs)
> 
> malloc shall return NULL on error
> malloc shall set errno to ENOMEM on error

It seems that the BSD versions do have ways to generate messages
depending on environment or global variables.  Otherwise, yes,
errno is the usual unix way to return error codes, and perror()
to print them out.
 
> So anyone arguing a *useful* error code is returned is onto a loser.
> Such is UNIX. RT11 and RSX had their equivalents of errno (the JSW and
> the DSW respectively) but the user experience was rather more
> friendly.

I remember finding out once that the VMS version of exit() 
prints out a message based on the return value.  I don't 
remember the message, but the one for 8 was surprising when
I wasn't expecting it.
 
> Unix has no "Error Messages and Recovery Procedures" manual, and for
> most applications errno.h might as well not exist either. How many
> different UNIX error messages from different applications can you get
> for the one simple error where RT11, RSX, and VMS would all
> consistently have said something clear and simple like "file not
> found" (or local language equivalent if applicable)?

Not so many, as they all have to be through values of errno.

-- glen

"John Wallace" <johnwallace4@yahoo.co.uk> wrote in message 
news:a7ed3d76-789f-49e6-b459-e6da503edcf2@a36g2000yqc.googlegroups.com...
On Oct 4, 11:13 pm, Bob Eager <rd...@spamcop.net> wrote:

>>>
Fred may not be 100% accurate in saying no error code is returned by
malloc(), but he is *right* about the Unix approach to error reporting
in general. Windows (with GetLastError()) is no better either.
>>>

For some reason, my newsreader isn't quoting your response.

Thanks.  Essentially because certain routines didn't include an error 
return, they created the errno hack.  Quite a hack... a global symbol.

I can defend things like C not having a general string type if C is viewed 
as a assembler replacement as opposed to a high-level language (although a 
counted string would make more sense in general than null terminated).

The language decision led to OS API decisions.

I've been a programmer for a living since around 1977/78.  The bulk of the 
code I've written in those 30+ years has been C.  I'd rather write C than 
write BLISS.  But you have to be blind to ignore the programming abuse I've 
seen in C and in the library interfaces.  You have to work hard to find a 
way to cause some of the common errors you see in C in other languages.

In article <yKKdnRFCD7DMtDfRnZ2dnUVZ_sOdnZ2d@giganews.com>, "Richard B. Gilbert"
<rgilbert88@comcast.net> writes:
> On 10/4/2010 10:49 AM, Michael Kraemer wrote:
> > In article<xrGdnah-dpDgQzTRnZ2dnUVZ_uudnZ2d@giganews.com>, "Richard B.
> Gilbert"
> > <rgilbert88@comcast.net>  writes:

> 
> If  a web browser is so urgent, why has no one ported Mozilla Firefox to 
> VMS?  And how about the Thunderbird mail/news client?  Could it be that 
> nobody cares any longer?

but, according to the links Mr. Maher was so kind to provide,
it *has* been ported. 
And this is what some people here are bitching about.
VMS is about the first platform I've heard people complain
that the vendor gives them more applications.

> 
> I don't miss having a browser on VMS, I can always use the one on my PC.
> Ditto for mail/news and the Thunderbird mail client

This may satisfy you and a few others, but it
isn't necessarily true for all VMS customers.

In article <8guoccF6t8U1@mid.individual.net>, billg999@cs.uofs.edu (Bill
Gunshannon) writes:
> 
> When are we going to put this myth to sleep.  Unix didn't invent null
> terminated strings.  DEC was perfectly happy with them as well.  All
> of the non-Unix PDP-11 OSes had them 

exactly.
The first time I encountered them (in a former life) 
was on a PDP-11 under RSX-11-something.
It seemed to be the standard way to store varying length strings, IIRC.
Quite efficient IMHO for an architecture which maxes out at 16 bits,
where a 2-byte length prefix might be considered as a waste of resources
compared to a 1-byte trailing null byte. 

> and no matter how hard people here
> try to deny it, so did VMS, from the very beginning (and, I'll bet it's
> still there in MACRO for anyone who wants to use it!)

In article <i8dak6$5h9$1@usenet01.boi.hp.com>, "FredK" <fred.nospam@dec.com>
writes:
> 
> 
> Bill, I didn't mean to imply UNIX or C "invented" zwro terminated strings. 
> C made it the language standard, UNIX enshrined it in it's interfaces. 
> Despite .ASCIZ - find it used as part of the calling standard.  It was 
> stupidity on a grand scale. 

And what's wrong with that?

In article <i8dd94$6ps$1@usenet01.boi.hp.com>, "FredK" <fred.nospam@dec.com>
writes:
> 
> "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message 
> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
> > I ask this naively. What is wrong with malloc ?  You give it a size, it
> > returns a pointer to a memory block allocated to you of that size.
> >
> 
> It also doesn't return an error code.

errno and the result pointer,
reflecting the two things which can go wrong:
an invalid number of bytes specified on input
and not enough memory available.
Do I miss something?

In article <a7ed3d76-789f-49e6-b459-e6da503edcf2@a36g2000yqc.googlegroups.com>,
John Wallace <johnwallace4@yahoo.co.uk> writes:

> 
> Fred may not be 100% accurate in saying no error code is returned by
> malloc(), but he is *right* about the Unix approach to error reporting
> in general. Windows (with GetLastError()) is no better either.
> 
> Find a unix and say
> man 3 malloc
> 
> Here is what it says about what happens on error (my summary, and
> consistent with various formal UNIX specs)
> 
> malloc shall return NULL on error
> malloc shall set errno to ENOMEM on error
> 
> So anyone arguing a *useful* error code is returned is onto a loser.
> Such is UNIX. 

BS.
Firstly my (but not necessarily everybody's) preference
would be calloc() rather than malloc() because it resets
the buffer to zero rather than rendering garbage
(with the intended side-effect that it is tested whether
the requested memory is really available),
and secondly this function gives an additional errno code
when one specifies a buffer length of zero, for example.
And what more information do you want from a memory allocator?
Either the RAM is there or it isn't.

> RT11 and RSX had their equivalents of errno (the JSW and
> the DSW respectively) but the user experience was rather more
> friendly.
> 
> Unix has no "Error Messages and Recovery Procedures" manual, and for
> most applications errno.h might as well not exist either. 

And just why would that be a drawback of Unix 
if the apps don't use what's there?

> How many
> different UNIX error messages from different applications can you get
> for the one simple error where RT11, RSX, and VMS would all
> consistently have said something clear and simple like "file not
> found" (or local language equivalent if applicable)?

That's the apps' fault, not Unix'.
All Unix' I've encountered in the past two decades
have perror() and strerror(), giving you the clear and simple
messages like "no such file" etc.
And they're literally the same across different flavours.

JF Mezei wrote:
> smithfarm wrote:
> 
>> On the other hand, if VMS is ever to gain any traction as a viable "real 
>> world" alternative to Windows and Linux, there is a certain minimal (and 
>> shifting) feature set that it must support.
> 
> Rewind 10-12 years: Enters Steve Jobs to rescue Apple. He job was to
> revamp the Macintosh product line and bring it back to life.  He used
> "vision" to convince the board to spend the big bucks to redevelop the OS.
> 
> Bringing VMS back to life as a server OS would take less effort than it
> took Apple to resurrect itself. Port it to the 64 bit 8086 and bring in
> modern stuff like IPsec and true IPv6 support.
> 
> Porting the browser is in fact fairly smart because it also forced the
> porting of much middleware such as php, perl and a buch of others that
> are used to build the Mozilla product. But it must remain a continuous
> effort to ensure that VMS has up-to-date middleware.
> 
> Also, if you need another OS to manage a server OS, you might as well
> use that other OS as your server OS.
> 
> 
> At this point in time we have to be realistic as painful as it may be.
> HP has had 10 years to change its mind about VMS.  It hasn't. It won't.
> 
> With VMS management and engineering gone, there is nobody to fight for
> VMS at HP.  And when the new guy inquires about HP's products, he will
> probably be told that it is too late to revive VMS and that the
> remaining VMS customers are happy with VMS being on maintenance mode
> with limited new development. There will be nobody to counter this.
> 

Why doesn't HP just put VMS up for sale?

> Existing mamagement like Livermore will arrange for the new guy to meet
> customers whose opinion of VMS matches what HP's policy on VMS is.
> 
> I don't know why anyone would still have hopes that HP would start a
> major turn around for VMS.

It amazes me why HP bought VMS and turn around and deny its existence to 
the world.  Do they get a tax write off or something?

Michael Kraemer wrote:
> In article <4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com>, JF Mezei
> <jfmezei.spamnot@vaxination.ca> writes:
>> Also, if you need another OS to manage a server OS, you might as well
>> use that other OS as your server OS.
>>
> 
> exactly.
> Moreover, to come up with another of those popular
> car analogies: what some people suggest here is
> to sell a car restricted to engine and tank ("server-centric").
> If you want to actually drive it,
> you'd have to buy/use a second one,
> which will give you seats, tires and a driving wheel,
> so you can hook up the first one.
> I think most, if not all customers would just
> buy the second, complete one, even if it 
> has a "weaker" (by whatever measure) engine
> than the first one.

You've got a good point.
Look at cummins diesel plant.  They don't make complete trucks, but they 
do sell their engines to other companies that do.  One would thing that 
HP could sell their engine to others that would complete the product 
offering.

But as I asked JF, why doesn't HP just sell VMS to someone else if they 
continue to hide it from the world on their web site?  Something just 
doesn't add up here.

FredK wrote:
> "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message 
> news:4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com...
>> smithfarm wrote:
> 
>> Bringing VMS back to life as a server OS would take less effort than it
>> took Apple to resurrect itself. Port it to the 64 bit 8086 and bring in
>> modern stuff like IPsec and true IPv6 support.
>>
> 
> I've said this 10000 times, and I'll repeat it 10000 more.  Porting VMS to 
> x86-64 for any and all of it's benefits - will make NO DIFFERENCE.  Or 
> rather, the difference would be so marginal that it would have no real 
> meaning.
> 
> Nobody really cares what OS is under the hood on Apple.  What they care 
> about are the applications that run on it.  They also remain a tiny niche of 
> the desktop market.  Nor, may I say are there computers what is leading 
> Apple's business anymore.  iPhones and iPods are the lifeblood of Apple.
> 

Well, I cared.  I have purchased a few Apple OS X machines and found no 
need for any anti-virus products.  Been using them for seven years now 
without any problems.

Now, VMS could very well be marketed the same way Apple found a way to 
market a better OS than OS 9.  All the customer needs is a no hassle 
easy to use product that is robust and secure... and VMS from my 
experience is quite a bit more secure than any UNIX.  (Which OS X is, 
currently).

FredK wrote:
> "smithfarm" <presnypreklad@gmail.com> wrote in message 
> news:i8d27r$2gh$1@speranza.aioe.org...
>>> I've said this 10000 times, and I'll repeat it 10000 more.  Porting VMS 
>>> to x86-64 for any and all of it's benefits - will make NO DIFFERENCE.  Or 
>>> rather, the difference would be so marginal that it would have no real 
>>> meaning.
>>>
>>> Nobody really cares what OS is under the hood on Apple.  What they care 
>>> about are the applications that run on it.  They also remain a tiny niche 
>>> of the desktop market.  Nor, may I say are there computers what is 
>>> leading Apple's business anymore.  iPhones and iPods are the lifeblood of 
>>> Apple.
>> I'm the last one who would argue with you, Fred.
>>
>> This begs a hypothetical question (I apologize in advance if it's been 
>> posed here a million times before): VMS is touted for its stability and 
>> security. If, in some parallel universe, VMS were the dominant OS running 
>> on 90% of end-user machines, would the security situation really be any 
>> better? Would there be fewer (and smaller) botnets, etc., less port 
>> scanning, less spam, etc.?
> 
> IMHO - security got out-of-control when it became possible to send methods 
> as opposed to data for useful purposes.  When it became possible for a 
> email, or a web site to execute code without the explicit control of the 
> user - the gates were wide open.
> 
> The PC for most people is a tool they use, and that they don't understand - 
> which is as it should be - you should not need to know how the engine in 
> your car works to drive one.  But that also means security settings will be 
> beyond most users to understand or control.  If a user gets a pop up that 
> says "Application foo is trying to read bar, do you want to allow it" - they 
> have no idea what to do.  What they want is to do all these useful and cool 
> things, which as a side effect introduces high risks.
> 
> From an OS perspective VMS is safer than UNIX-like OSes for many aspects of 
> operation because of roots in it's design - part of which was language and 
> calling conventions.  VMS would never have chosen "malloc" as a memory 
> allocation API.  Or depended on zero termination for strings.  The fact that 
> we now have a lot of C code makes us vulnerable - but the good news is that 
> the OS calling conventions haven't changed - so the effect tends to be more 
> around application crashers.  There is little we can do about TCPIP - it 
> won, OSI lost.  We have to live and try to deal with the problems that come 
> along with it.  If VMS had won, TCPIP would not have.
> 

If I recall, too many customers were needing TCPIP from DEC to connect 
to other platforms.  Back in the 80s there were a lot of different 
protocols being offered and was a bit confusing as to which to pick.
You may have some better history on this topic than I can recall tho.

> Apple with a UNIX-like OS under the hood has fewer issues than windows.

Correction:  OS X is unix certified.

> VMS 
> has less than UNIX-like OS's.  That isn't to say that we don't have security 
> bugs.  I've fixed at least one within the last decade.  But they are 
> generally less likely - and there are fewer eyes and hands trying to find 
> them and exploit them.  If VMS had beaten UNIX - then more would be found. 
> But the question is if VMS had beaten UNIX - would C/C++ languages be the 
> dominant languages?  Or Pascal and PL/I (Fortran, Bliss, Ada, ...)?
> 

Knowing DEC back then... it would be Fortran.  And todays Fortran 
standards are pretty good.  It is the object oriented aspect that would 
have to be considered.  I find Objective-C to quite good at this, but 
then also can get quite obscure as well to someone else reading the 
code, but also very elegant.

Michael Kraemer wrote:
> In article <8guoccF6t8U1@mid.individual.net>, billg999@cs.uofs.edu (Bill
> Gunshannon) writes:
>> When are we going to put this myth to sleep.  Unix didn't invent null
>> terminated strings.  DEC was perfectly happy with them as well.  All
>> of the non-Unix PDP-11 OSes had them 
> 
> exactly.
> The first time I encountered them (in a former life) 
> was on a PDP-11 under RSX-11-something.
> It seemed to be the standard way to store varying length strings, IIRC.
> Quite efficient IMHO for an architecture which maxes out at 16 bits,
> where a 2-byte length prefix might be considered as a waste of resources
> compared to a 1-byte trailing null byte. 

I remember those days well.  Storage and memory was at a big premium...
something the current generation of giga-byter programmers don't 
understand from back then.  When I mean big premium, I'm talking about a 
few hundred thousand dollars for a couple of 500Mb hard drives and 32mb 
of memory.
Today that same resource is just chump change.

JF Mezei wrote:
> FredK wrote:
> 
>> Don't get me wrong in thinking that I would not like to see a x86-64 VMS 
>> box - I just do not believe it is the magic bullet you apparently think it 
>> is.
> 
> Porting VMS to the 8086 is not a magic bullet. Changing HP's mind on VMS
> is the magic bullet.
> 
> Porting to the 8086 would be a sign that HP is changing its mind about
> VMS and would likely be accompanied with other steps such as allowing
> marketing of VMS etc.
> 
> When the "HP is killing VMS" noise gets too loud, HP will simply take a
> token action (such as one time marketing) to quiet that noise. This has
> happened since the palmer era. So token actions are no longer sufficient
> to convince people of a real change.
> 
> The port of VMS to 8086 would be a substantial sign of a commitment one
> that would not be short lived.

I've also read on CNET news that HP isn't innovative with software.
Considering this point, is it any wonder?
VMS port to generic intel platforms is no big deal.  All it takes from 
HP is some real UI glitz and a very easy to use platform.  But it would 
take another Steve Jobs to do that.  No one I know of his calibre works 
at HP.

GreyCloud wrote:

> It amazes me why HP bought VMS and turn around and deny its existence to 
> the world.  Do they get a tax write off or something?

HP didn't buy VMS. They bought Compaq's PC market share, DEC's service
contracts, and the disk array business. One could argue that they also
bought the iPaq product line, but like with VMS, they did very little
with it, and now completely abandonned it when they bought Palm.

GreyCloud wrote:

> But as I asked JF, why doesn't HP just sell VMS to someone else if they 
> continue to hide it from the world on their web site?  Something just 
> doesn't add up here.


I am not sure whether they consider VMS to have intellectual property
worth keeping and that shouldn't be given to a competitor,

OR

HP management were told that VMS, while having no resale value, still
generates enough support contract revenues to continue to hold on to it
and pretend it is still being developed.

On 4 okt, 22:25, JF Mezei <jfmezei.spam...@vaxination.ca> wrote:
> FredK wrote:
> > I've said this 10000 times, and I'll repeat it 10000 more. =A0Porting V=
MS to
> > x86-64 for any and all of it's benefits - will make NO DIFFERENCE. =A0O=
r
> > rather, the difference would be so marginal that it would have no real
> > meaning.
>
> So Mr FredK, as one of the few remaining real VMS engineers left at HP,
> if the new CEO were to ask your opinion, this is what you would tell him?
>
> You would rather see VMS restricted to an unwanted IA64 platform whose
> future is uncertain and performance now below par compared to the 8086 ?


JF,
I'm afraid that Fred's reply will be something like "VMS has been
neglected too long and thus too far away from being rescued by new
hardware".
The new CEO doesn't know VMS from AOS and will easily sign an order to
close down VMS development. Fewer customers means reduced project
funding.
If your boss reads about the build quality of VMS 8.4 on c.o.v. it
will just make it easier to move to a reliable platform, FCOL.
Hans

On 4 okt, 22:35, JF Mezei <jfmezei.spam...@vaxination.ca> wrote:
> FredK wrote:
> > calling conventions. =A0VMS would never have chosen "malloc" as a memor=
y
> > allocation API.
>
> I ask this naively. What is wrong with malloc ? =A0You give it a size, it
> returns a pointer to a memory block allocated to you of that size.
>
> > Apple with a UNIX-like OS under the hood has fewer issues than windows.=
 =A0VMS
> > has less than UNIX-like OS's. =A0That isn't to say that we don't have s=
ecurity
> > bugs.
>
> One metric used these days is the number of days between the discovery
> of a bug and a patch being made available. (aka: length of vulnerability)=
..
>
> For that metric, VMS is terrible. Thankfully, VMS is so unimportant that
> its security bugs don't get reported to CERT or other widely publiscised
> mediums.
>
> > But the question is if VMS had beaten UNIX - would C/C++ languages be t=
he
> > dominant languages? =A0Or Pascal and PL/I (Fortran, Bliss, Ada, ...)?
>
> The world would be all in BLISS and DIBOL :-) :-) :-)

VMS (and DEC) has its influences in many areas of IT but I never
considered DEC as a language development company.
BLISS is probably the best example, and it is about as appealing to me
as C is (I learned to program in Algol, in the Dijkstra school).
I don't think Ada would have made it as a system programming language,
but PL/1 might have done that. Pascal, like Algol60, is too simple.
It might have been extended, like Burroughs did with Algol, to a
language in which you could write operating systems and compilers.
Hans

In article <7c5642b2-7d20-4d89-8a2e-b052f9fe02a8@26g2000yqv.googlegroups.com>, H
Vlems <hvlems@freenet.de> writes:
 
> VMS (and DEC) has its influences in many areas of IT 

almost forgotten but probably most influential until now:
X Window and Ethernet.

> but I never
> considered DEC as a language development company.

it was a hardware company in the first place.

> BLISS is probably the best example, and it is about as appealing to me
> as C is (I learned to program in Algol, in the Dijkstra school).
> I don't think Ada would have made it as a system programming language,
> but PL/1 might have done that. 

Way too heavy weight. Some platforms never saw a PL/I compiler,
and the most popular ones only when it was much too late
and C/C++ had filled the void.

> Pascal, like Algol60, is too simple.
> It might have been extended, like Burroughs did with Algol, to a
> language in which you could write operating systems and compilers.

Such a language has to be simple and readily available
on whatever piece of silicon out there, which leaves - C.

On 2010-10-05 07:04, GreyCloud wrote:

> But as I asked JF, why doesn't HP just sell VMS to someone else if they
> continue to hide it from the world on their web site?

Do they ?

If I go to www.hp.com (and beeing redirectec to the swedish HP site),
I can look under "Servers" and have OpenVMS listed together
with the other OS'es. Or click "Software" and have OpenVMS listed
on the next page again together with all the other OS'es.

In what way is OpenVMS hided from the world any more then any of
the other operating systems ?

In article <YdadnaCKHrxJLzfRnZ2dnUVZ_q6dnZ2d@bresnan.com>, GreyCloud
<mist@cumulus.com> writes:

> 
> Well, I cared.  I have purchased a few Apple OS X machines and found no 
> need for any anti-virus products.  Been using them for seven years now 
> without any problems.
> 
> Now, VMS could very well be marketed the same way Apple found a way to 
> market a better OS than OS 9.  All the customer needs is a no hassle 
> easy to use product that is robust and secure... and VMS from my 
> experience is quite a bit more secure than any UNIX.  (Which OS X is, 
> currently).

Where's the logic?
If OS X (a BSD derivative) already gives you zero problems,
how would VMS (being allegedly more "secure") improve on that?
Does a negative number of problems exist?

Moreover, the OS X analogy is incorrect.
Jobs relegated the old OS 9 to a compatibility box
and put a completely new (Unix) OS under the hood,
plus some new fancy apps on top.
Translated to VMS this would mean to dump its traditional kernel
and replace it by some Unix. Somehow I suspect that this is
not what VMS bigots would want.

In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>, GreyCloud
<mist@cumulus.com> writes:

> But as I asked JF, why doesn't HP just sell VMS to someone else 

Why should they?
They don't have a problem with it.
Contrary to common belief here the world doesn't have problems
without it.
So why change the status quo?

> if they 
> continue to hide it from the world on their web site? 

Do they?
So far you would find all VMS related info on their respective pages.
Of course not on the front pages, no wonder
since 99+x% of HP customers are interested in PCs and ink
rather than nerd stuff such as VMS. 

> Something just 
> doesn't add up here.

On Mon, 2010-10-04 at 04:18 -0400, JF Mezei wrote:

> I don't know why anyone would still have hopes that HP would start a
> major turn around for VMS.=20

Can't someone just step up and buy the rights to OpenVMS from HP,
become an independent vendor, and live off the sales and support
revenue stream?=20

Then port like mad to other architectures such as x86 and AMD64?=20

Or even open source the whole thing.=20
--=20
Tactical Nuclear Kittens

There *is* enough money already!

It's just that it's *still* being siphoned into the personal pet projects of 
the scum that has survived 30 years without writing a single line of VMS or 
layered product code :-(

Regards Roger the cabin boy.

"Captain Pugwash" <alex.buell@munted.org.uk> wrote in message 
news:1ucrn7-pib.ln1@nntp.local.net...
On Mon, 2010-10-04 at 04:18 -0400, JF Mezei wrote:

> I don't know why anyone would still have hopes that HP would start a
> major turn around for VMS.

Can't someone just step up and buy the rights to OpenVMS from HP,
become an independent vendor, and live off the sales and support
revenue stream?

Then port like mad to other architectures such as x86 and AMD64?

Or even open source the whole thing.
-- 
Tactical Nuclear Kittens

On Tue, 2010-10-05 at 19:53 +0800, Richard Maher wrote:

> There *is* enough money already!
>=20
> It's just that it's *still* being siphoned into the personal pet projects=
 of=20
> the scum that has survived 30 years without writing a single line of VMS =
or=20
> layered product code :-(

Just out of interest just how many organisations out there still use
OpenVMS or VMS?=20
--=20
Tactical Nuclear Kittens

In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>,
	GreyCloud <mist@cumulus.com> writes:
> Michael Kraemer wrote:
>> In article <4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com>, JF Mezei
>> <jfmezei.spamnot@vaxination.ca> writes:
>>> Also, if you need another OS to manage a server OS, you might as well
>>> use that other OS as your server OS.
>>>
>> 
>> exactly.
>> Moreover, to come up with another of those popular
>> car analogies: what some people suggest here is
>> to sell a car restricted to engine and tank ("server-centric").
>> If you want to actually drive it,
>> you'd have to buy/use a second one,
>> which will give you seats, tires and a driving wheel,
>> so you can hook up the first one.
>> I think most, if not all customers would just
>> buy the second, complete one, even if it 
>> has a "weaker" (by whatever measure) engine
>> than the first one.
> 
> You've got a good point.
> Look at cummins diesel plant.  They don't make complete trucks, but they 
> do sell their engines to other companies that do.  One would thing that 
> HP could sell their engine to others that would complete the product 
> offering.
> 
> But as I asked JF, why doesn't HP just sell VMS to someone else if they 
> continue to hide it from the world on their web site?  Something just 
> doesn't add up here.

Probably because the perciened cost of selling the product line is greater
than any percieved benefit.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8dak6$5h9$1@usenet01.boi.hp.com>,
	"FredK" <fred.nospam@dec.com> writes:
> 
> "Bill Gunshannon" <billg999@cs.uofs.edu> wrote in message 
> news:8guoccF6t8U1@mid.individual.net...
>> In article <i8d3ae$1ak$1@usenet01.boi.hp.com>,
>> "FredK" <fred.nospam@dec.com> writes:
>>>
>>> From an OS perspective VMS is safer than UNIX-like OSes for many aspects 
>>> of
>>> operation because of roots in it's design - part of which was language 
>>> and
>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>> allocation API.  Or depended on zero termination for strings.
>>
>> When are we going to put this myth to sleep.  Unix didn't invent null
>> terminated strings.  DEC was perfectly happy with them as well.  All
>> of the non-Unix PDP-11 OSes had them and no matter how hard people here
>> try to deny it, so did VMS, from the very beginning (and, I'll bet it's
>> still there in MACRO for anyone who wants to use it!)
>>
> 
> Bill, I didn't mean to imply UNIX or C "invented" zwro terminated strings. 

Not singling you out.  It is just a common thread here where people
seem to ignore the fact that Unix/C merely picked up on a method that
was pretty much standard accross the industry.

> C made it the language standard, 

De facto made it a standard.  To the best of my knowledge C does not have
a string data type at all. 

>                                   UNIX enshrined it in it's interfaces. 

You mean like the example I posted for VMS?  Use made it what it is and
people are very unlikely to make major changes that break lots of already
written software.  That is what caused all the two year dates more than a
decade after punched-cards went away and keeps time_t too small and even
keeps IPV6 out of the mainstream (I first used IPV6 more than 10 years
ago and we actually had (have?) an address allocation).

> Despite .ASCIZ - find it used as part of the calling standard.  

The fact that it is still included as a part of the language tells me
is is, in fact, a "part of the calling standar" on VMS.  Just because
people choose to use something better doesn't remove it from the
standard.

>                                                                  It was 
> stupidity on a grand scale. 
 
Hindsight is always 20/20.  It served a purpose and worked very well for
a long time.  It is still usable and perfectly safe depending on the
programmer.   And, in some environments it is perfectly safe regardless.
I would foresee no problem using .ASCIZ cwunder RT-11, RSTS, RSX or under
VMS.  Of course, ~i have no problem with using it under Unix, but that's
me.  :-)

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8dd94$6ps$1@usenet01.boi.hp.com>,
	"FredK" <fred.nospam@dec.com> writes:
> 
> "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message 
> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>> FredK wrote:
>>
>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>> allocation API.
>>
>> I ask this naively. What is wrong with malloc ?  You give it a size, it
>> returns a pointer to a memory block allocated to you of that size.
>>
> 
> It also doesn't return an error code.

It most certainly does.  The only available failure is if therre is not enough
memory avaialable to meet the request.  It then returns NULL and sets errno
to ENOMEM - Not enough Memory.

What other errors are you expecting?

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8dikb$9tj$1@usenet01.boi.hp.com>,
	"FredK" <fred.nospam@dec.com> writes:
> 
> "Bob Eager" <rde42@spamcop.net> wrote in message 
> news:8gv1rnFefcU5@mid.individual.net...
>> On Mon, 04 Oct 2010 17:01:02 -0400, FredK wrote:
>>
>>> "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message
>>> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>>>> FredK wrote:
>>>>
>>>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>>>> allocation API.
>>>>
>>>> I ask this naively. What is wrong with malloc ?  You give it a size, it
>>>> returns a pointer to a memory block allocated to you of that size.
>>>>
>>>>
>>> It also doesn't return an error code.
>>
>> If it complies with the C standard, it does.
>>
> 
> foo = malloc(size);
> 
> The only "error" return is NULL.   

Wrong.  The NULL is not an error return it is how much memory was allocated.

>                                   Returning NULL provides me with zero 
> information that can be either used to correct the problem, or instruct the 
> user how to correct the problem - other than guessing.

Unless, you decide to read tha man page which tells you to check errno.

>                                                          So to be more long 
> winded, if there can be more than one possible error failure, a function 
> should return a seperate error code.

There is only one possible failure of malloc().  No memory to allocate.
And it does return a separate error code even though there is only one.
I can think of one possible reason why malloc() would return NULL and
have it not be an error.  if you asked for NULL bytes.  Otherwise the
only cause for a return of NULL would be not enough memory available to
meet the request and that is handled perfectly well as documented.

> 
> malloc was simply an example of the generic aversion to error handling, 
> followed by kludges to provide errors.
 
And you were wrong.

bill 
 
 

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <8gv5okFefcU7@mid.individual.net>,
	Bob Eager <rde42@spamcop.net> writes:
> On Mon, 04 Oct 2010 18:32:21 -0400, FredK wrote:
> 
>> "Bob Eager" <rde42@spamcop.net> wrote in message
>> news:8gv1rnFefcU5@mid.individual.net...
>>> On Mon, 04 Oct 2010 17:01:02 -0400, FredK wrote:
>>>
>>>> "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message
>>>> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>>>>> FredK wrote:
>>>>>
>>>>>> calling conventions.  VMS would never have chosen "malloc" as a
>>>>>> memory allocation API.
>>>>>
>>>>> I ask this naively. What is wrong with malloc ?  You give it a size,
>>>>> it returns a pointer to a memory block allocated to you of that size.
>>>>>
>>>>>
>>>> It also doesn't return an error code.
>>>
>>> If it complies with the C standard, it does.
>>>
>>>
>> foo = malloc(size);
>> 
>> The only "error" return is NULL.   Returning NULL provides me with zero
>> information that can be either used to correct the problem, or instruct
>> the user how to correct the problem - other than guessing.  So to be
>> more long winded, if there can be more than one possible error failure,
>> a function should return a seperate error code.
>> 
>> malloc was simply an example of the generic aversion to error handling,
>> followed by kludges to provide errors.
> 
> Or an example of a distorted hate campaign...
> 
> If you are determined to find fault, then  of course you will do so.
> 
 
Even when a simple read of any manpage would show it was just plain wrong. 

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <8h0oacF3bgU5@mid.individual.net>, billg999@cs.uofs.edu (Bill
Gunshannon) writes:
> In article <i8dd94$6ps$1@usenet01.boi.hp.com>,
> 	"FredK" <fred.nospam@dec.com> writes:
> > 
> > "JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message 
> > news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
> >> FredK wrote:
> >>
> >>> calling conventions.  VMS would never have chosen "malloc" as a memory
> >>> allocation API.
> >>
> >> I ask this naively. What is wrong with malloc ?  You give it a size, it
> >> returns a pointer to a memory block allocated to you of that size.
> >>
> > 
> > It also doesn't return an error code.
> 
> It most certainly does.  The only available failure is if therre is not
> enough
> memory avaialable to meet the request.  It then returns NULL and sets errno
> to ENOMEM - Not enough Memory.
> 
> What other errors are you expecting?
> 

well, "man malloc" for example on AIX tells me:

<quote>
Error Codes

When the memory allocation subroutines are unsuccessful, the global variable
errno may be set to the following values:

EINVAL Indicates a call has requested 0 bytes.

ENOMEM Indicates that not enough storage space was available.
</quote>

where the EINVAL might be useful in the debugging phase.

On 5 okt, 10:34, m.krae...@gsi.de (Michael Kraemer) wrote:
> In article <7c5642b2-7d20-4d89-8a2e-b052f9fe0...@26g2000yqv.googlegroups.com>, H
>
> Vlems <hvl...@freenet.de> writes:
> > VMS (and DEC) has its influences in many areas of IT
>
> almost forgotten but probably most influential until now:
> X Window and Ethernet.
>
> > but I never
> > considered DEC as a language development company.
>
> it was a hardware company in the first place.
>
> > BLISS is probably the best example, and it is about as appealing to me
> > as C is (I learned to program in Algol, in the Dijkstra school).
> > I don't think Ada would have made it as a system programming language,
> > but PL/1 might have done that.
>
> Way too heavy weight. Some platforms never saw a PL/I compiler,
> and the most popular ones only when it was much too late
> and C/C++ had filled the void.
>
> > Pascal, like Algol60, is too simple.
> > It might have been extended, like Burroughs did with Algol, to a
> > language in which you could write operating systems and compilers.
>
> Such a language has to be simple and readily available
> on whatever piece of silicon out there, which leaves - C.

It's not easy to summarize what DEC achieved in a couple of words.
First and foremost a hardware company
but DEC also built fine software products.Their OSes were DEC's own
design. DEC also had an excellent
reputation for building compilers, its members continue to do so but
now for Intel and others.
And no, I didn't forget about ethernet. The point was whether DEC did
design a programming language that
had as much influence as its other products did. And even though BLISS
is a highly protable product, it never
took the world. Neither did Burroughs Extended Algol. Portability
isn't the only issue otherwise linux would have
been written in COBOL.
PL/I is a rich language with lots of functionality to write system
software in. It may be too rich, that's probably why
subsets came into existence for exactly that purpose. There are quite
a lot of proprietary derivatives, which probably
is the reason why they never took off.
And no, C is not always easy to implement, ask Unisys.
Hans

In article <c7709661-68a9-4722-9106-bc3842706467@t20g2000yqa.googlegroups.com>,
H Vlems <hvlems@freenet.de> writes:
> 
> It's not easy to summarize what DEC achieved in a couple of words.
> First and foremost a hardware company
> but DEC also built fine software products.Their OSes were DEC's own
> design. DEC also had an excellent
> reputation for building compilers, 

Not for me. They had just as many bugs
as the competitor's, if not more,
in particular on DEC's RISC machines.
Could be, of course, that they just weren't given
enough time to mature, unlike on VAX.

> its members continue to do so but
> now for Intel and others.
> And no, I didn't forget about ethernet. The point was whether DEC did
> design a programming language that
> had as much influence as its other products did. And even though BLISS
> is a highly protable product, 

since nobody uses it outside of the VMS niche,
this is a bit hard to verify.

> it never
> took the world. Neither did Burroughs Extended Algol. Portability
> isn't the only issue otherwise linux would have
> been written in COBOL.

I can't see why Cobol should be more portable than C.

> PL/I is a rich language with lots of functionality to write system
> software in. 

my experience from a former life is just the opposite.
It was good for high level stuff like formatted I/O,
but as soon as things came closer to OS and hardware,
assembly language was a much better choice. 

> It may be too rich, that's probably why
> subsets came into existence for exactly that purpose. 

and there goes portability.
If supported subsets differ from one compiler 
to another, you loose. 

> There are quite
> a lot of proprietary derivatives, which probably
> is the reason why they never took off.

Exactly. There are too many constructs defined in the 
language itself, making it impossible to even compile
certain PL/I sources on a different platform.
For example, unless the standard has evolved significantly
in the past 20 years, stuff like

DCL FILE ENV(vsam);

won't compile properly on anything not supporting that IBM stuff.
On C one would use sth like

fp = fopen( filename, "type=vsam" );

which would at least compile and one can hide
platform dependent stuff in literals and/or ifdef's.


> And no, C is not always easy to implement, ask Unisys.

who's Unisys?
And, if they aren't even able to implement a language
as simple as C, should I want to know? :-)

In article <a7ed3d76-789f-49e6-b459-e6da503edcf2@a36g2000yqc.googlegroups.com>, John Wallace <johnwallace4@yahoo.co.uk> writes:
> 
> Unix has no "Error Messages and Recovery Procedures" manual, and for
> most applications errno.h might as well not exist either. How many
> different UNIX error messages from different applications can you get
> for the one simple error where RT11, RSX, and VMS would all
> consistently have said something clear and simple like "file not
> found" (or local language equivalent if applicable)?

   Reminds me of Fortran programming on a VAX.  If you captured an
   error opening a file in Fortran, it was mapped to something 
   like "couldn't open the file".  Now that's real helpfull, but that's
   the Fortran standard.

   If you didn't catch the error, the traceback started with the
   meaningless Fortran message, and continued into RMS messages and such
   until you knew what happened.

In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>, GreyCloud <mist@cumulus.com> writes:
> 
> But as I asked JF, why doesn't HP just sell VMS to someone else if they 
> continue to hide it from the world on their web site?  Something just 
> doesn't add up here.

   Why sell a product to a potential competitor that could then
   actually compete with you?

   There could be some way to sell lots of ink if you owned VMS.  8-)

In article <a7ed3d76-789f-49e6-b459-e6da503edcf2@a36g2000yqc.googlegroups.com>,
	John Wallace <johnwallace4@yahoo.co.uk> writes:
> On Oct 4, 11:13�pm, Bob Eager <rd...@spamcop.net> wrote:
>> On Mon, 04 Oct 2010 17:01:02 -0400, FredK wrote:
>> > "JF Mezei" <jfmezei.spam...@vaxination.ca> wrote in message
>> >news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>> >> FredK wrote:
>>
>> >>> calling conventions. �VMS would never have chosen "malloc" as a memory
>> >>> allocation API.
>>
>> >> I ask this naively. What is wrong with malloc ? �You give it a size, it
>> >> returns a pointer to a memory block allocated to you of that size.
>>
>> > It also doesn't return an error code.
>>
>> If it complies with the C standard, it does.
>>
>> --
>> Use the BIG mirror service in the UK:
>> �http://www.mirrorservice.org
>>
>> *lightning protection* - a w_tom conductor
> Fred may not be 100% accurate in saying no error code is returned by
> malloc(), but he is *right* about the Unix approach to error reporting
> in general. 

Guess it's a matter of opinion or more likely what you are most familiar
with.  I have no problem with the Unix way.

>             Windows (with GetLastError()) is no better either.

Have to agree with this.  I have never been able to decipher Windows
error messages. But then, could be the same as what I said above.  I
would bet that windows programmers have little if any problem with it.

> Find a unix and say
> man 3 malloc
> Here is what it says about what happens on error (my summary, and
> consistent with various formal UNIX specs)
> malloc shall return NULL on error
> malloc shall set errno to ENOMEM on error
> So anyone arguing a *useful* error code is returned is onto a loser.

Why?  The only cause of an error return from malloc(0 is if there isn't
enough memory to meet the request.  ENOMEM is a numeric value intended
to be interpreted and the programmer is free to return any human readable
message to the user he wishes if he can't programatically fix the problem.
ENOMEM is not intended for users.  If one actually looks at what WENOMEM
means it is simply: "Can't allocate memory".  How much cleare can you be?

And if you don't want to write your own message you merely call perror
and it will take care of it for you. 


> Such is UNIX. RT11 and RSX had their equivalents of errno (the JSW and
> the DSW respectively) but the user experience was rather more
> friendly.

How so?  the value of errno is never passed to the uiser unless the
programmer is an idiot.  The user either gets a program generated
message or the human readable system error message. 

Ya know.  People here should learn Unix before they try criticizing it.

> Unix has no "Error Messages and Recovery Procedures" manual, 

Doesn't need a whole manual.  it all fits in a couple of pages of the
online manuals.
Try "'man errno"and "man perror".

>                                                               and for
> most applications errno.h might as well not exist either. 

Aaare we talking applications or programmer interfaces?  I don't know
many users who call malloc() from the keyboard.

>                                                             How many
> different UNIX error messages from different applications can you get
> for the one simple error where RT11, RSX, and VMS would all
> consistently have said something clear and simple like "file not
> found" (or local language equivalent if applicable)?

If it is anything other than the one official error statement per errno
then it is the application's programmer doing it and has nothing to do
with Unix.   Unix provides perror() which returns the official Unix
explanation of errno.  Anything else is not Unix.  Or are you trying
to tell me that RT-11, RSX and VMS won't allow the programmer to return
their own custom messages for non-fatal errors?

bill


-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

On 10/5/2010 1:01 AM, GreyCloud wrote:
> JF Mezei wrote:
>> smithfarm wrote:
>>
>>> On the other hand, if VMS is ever to gain any traction as a viable
>>> "real world" alternative to Windows and Linux, there is a certain
>>> minimal (and shifting) feature set that it must support.
>>
>> Rewind 10-12 years: Enters Steve Jobs to rescue Apple. He job was to
>> revamp the Macintosh product line and bring it back to life. He used
>> "vision" to convince the board to spend the big bucks to redevelop the
>> OS.
>>
>> Bringing VMS back to life as a server OS would take less effort than it
>> took Apple to resurrect itself. Port it to the 64 bit 8086 and bring in
>> modern stuff like IPsec and true IPv6 support.
>>
>> Porting the browser is in fact fairly smart because it also forced the
>> porting of much middleware such as php, perl and a buch of others that
>> are used to build the Mozilla product. But it must remain a continuous
>> effort to ensure that VMS has up-to-date middleware.
>>
>> Also, if you need another OS to manage a server OS, you might as well
>> use that other OS as your server OS.
>>
>>
>> At this point in time we have to be realistic as painful as it may be.
>> HP has had 10 years to change its mind about VMS. It hasn't. It won't.
>>
>> With VMS management and engineering gone, there is nobody to fight for
>> VMS at HP. And when the new guy inquires about HP's products, he will
>> probably be told that it is too late to revive VMS and that the
>> remaining VMS customers are happy with VMS being on maintenance mode
>> with limited new development. There will be nobody to counter this.
>>
>
> Why doesn't HP just put VMS up for sale?
>

Who would buy it?  It comes with obligations to support the O/S for 
D.O.D. for the next eight or ten years.  Damned few will buy it if 
support is not available.

Most of us need little or no support.  We have the O/S, know the O/S and 
have used it for twenty or thirty years.  Give us the patches and we 
will keep going till hell freezes.  If worst comes to worst, some of our 
members have the skills to maintain the O/S.  What we don't have is the 
source code,  program logic manuals, and other tools needed to build
from source.

Our members grow old, retire, and eventually die.  Where are the new 
users to replace us to come from?

I think the war was lost back in the late 1980s and early 1990s when 
DEC failed/refused to compete in the workstation market.  The last 
twenty years or so have been a gallant rear guard action!

On 2010-10-05, Bill Gunshannon <billg999@cs.uofs.edu> wrote:
> In article <i8dak6$5h9$1@usenet01.boi.hp.com>,
> 	"FredK" <fred.nospam@dec.com> writes:
>
>> C made it the language standard, 
>
> De facto made it a standard.  To the best of my knowledge C does not have
> a string data type at all. 
>

Without getting involved in the rest of the discussion, I also believe C
made it a standard when null terminated strings became part of the core
C RTL (ie: strcpy and friends).

The core RTL functions are a attribute of the language and not the
underlying OS environment. It just so happens that on most (all?) Unix
type systems the library is supplied as part of the OS.

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980's technology to a 21st century world

On Oct 5, 5:15=A0pm, Simon Clubley <clubley@remove_me.eisner.decus.org-
Earth.UFP> wrote:
> On 2010-10-05, Bill Gunshannon <billg...@cs.uofs.edu> wrote:
>
> > In article <i8dak6$5h...@usenet01.boi.hp.com>,
> > =A0 =A0"FredK" <fred.nos...@dec.com> writes:
>
> >> C made it the language standard,
>
> > De facto made it a standard. =A0To the best of my knowledge C does not =
have
> > a string data type at all.
>
> Without getting involved in the rest of the discussion, I also believe C
> made it a standard when null terminated strings became part of the core
> C RTL (ie: strcpy and friends).
>
> The core RTL functions are a attribute of the language and not the
> underlying OS environment. It just so happens that on most (all?) Unix
> type systems the library is supplied as part of the OS.
>
> Simon.
>
> --
> Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
> Microsoft: Bringing you 1980's technology to a 21st century world

Null terminated strings, and the associated string manipulation
routines, were in the first edition of K+R. It's hard to see what
could be more "standard" than that, as far as original C goes, and
although there were some differences when ANSI C came out, nothing
directly relevant here changed. As you said, C is largely C,
regardless of OS. [Some limited exceptions apply; e.g. Whitesmith's C
for RSX didn't come with printf when I bought it. Soon fixed though.]

In article <i8drpb$mjs$1@speranza.aioe.org>,
	glen herrmannsfeldt <gah@ugcs.caltech.edu> writes:
> John Wallace <johnwallace4@yahoo.co.uk> wrote:
> (snip on malloc error messages)
>  
>> Fred may not be 100% accurate in saying no error code is returned by
>> malloc(), but he is *right* about the Unix approach to error reporting
>> in general. Windows (with GetLastError()) is no better either.
>> 
>> Here is what it says about what happens on error (my summary, and
>> consistent with various formal UNIX specs)
>> 
>> malloc shall return NULL on error
>> malloc shall set errno to ENOMEM on error
> 
> It seems that the BSD versions do have ways to generate messages
> depending on environment or global variables.  Otherwise, yes,
> errno is the usual unix way to return error codes, and perror()
> to print them out.

Which returns the  simple user level, human readable error message
like the "file not found" mentioned below.  Which doesn't preclude
the programmer returning any human readable message he might choose
to concoct.

>  
>> So anyone arguing a *useful* error code is returned is onto a loser.
>> Such is UNIX. RT11 and RSX had their equivalents of errno (the JSW and
>> the DSW respectively) but the user experience was rather more
>> friendly.
> 
> I remember finding out once that the VMS version of exit() 
> prints out a message based on the return value.  I don't 
> remember the message, but the one for 8 was surprising when
> I wasn't expecting it.

To be quite honest, I would be very surprised if VMS didn't actually
just return a number when an error occured, just like Unix does and
have another routine print the human readable error message.

>  
>> Unix has no "Error Messages and Recovery Procedures" manual, and for
>> most applications errno.h might as well not exist either. How many
>> different UNIX error messages from different applications can you get
>> for the one simple error where RT11, RSX, and VMS would all
>> consistently have said something clear and simple like "file not
>> found" (or local language equivalent if applicable)?
> 
> Not so many, as they all have to be through values of errno.

Say what!!  There are about 90 predefined and "'errno" is an int.
How many possible values does that make?

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8dq16$d9r$1@usenet01.boi.hp.com>,
	"FredK" <fred.nospam@dec.com> writes:
> 
> "John Wallace" <johnwallace4@yahoo.co.uk> wrote in message 
> news:a7ed3d76-789f-49e6-b459-e6da503edcf2@a36g2000yqc.googlegroups.com...
> On Oct 4, 11:13 pm, Bob Eager <rd...@spamcop.net> wrote:
> 
>>>>
> Fred may not be 100% accurate in saying no error code is returned by
> malloc(), but he is *right* about the Unix approach to error reporting
> in general. Windows (with GetLastError()) is no better either.
>>>>
> 
> For some reason, my newsreader isn't quoting your response.
> 
> Thanks.  Essentially because certain routines didn't include an error 
> return, they created the errno hack.  Quite a hack... a global symbol.

Matter of opinion.  It has served Unix programmers well for a lot longer
than VMS has even been around.

> 
> I can defend things like C not having a general string type if C is viewed 
> as a assembler replacement as opposed to a high-level language (although a 
> counted string would make more sense in general than null terminated).

If you want a counted string, create it.  That's one of the beauties of
C.  The basic types as might be defined at the hardware level already
exist and you are free to create any other user defined type you feel
necessary.  Maker a string what you want a string to be.  I have worked
with counted strings (ala UCSD Pascal) I saw no real advantage to them. 
And one very big disadvantage.  They come with a very finite fixed maximum
length.  If nothing else, that either keeps them very short or makes them
very non-portable.

> 
> The language decision led to OS API decisions.

How?  The language does not define a "string" type.  That is actually
what is in the OS API.  And there have been numerous oportunities to
change it.  No one showed any interest. (Or have we all forgotten things
line "Safe C" which was created back in the 70's and as a commercial
product, didn't fly.)

> 
> I've been a programmer for a living since around 1977/78.  

I first programmed around 1971 but it was shortly after you that I took
it up as a full-time job.

>                                                             The bulk of the 
> code I've written in those 30+ years has been C.  

While I have 9and still do) write a lot of C I have actually written 
programs in well over a dozen different languages many of which I have
long since forgotten.

>                                                   I'd rather write C than 
> write BLISS.  But you have to be blind to ignore the programming abuse I've 
> seen in C and in the library interfaces.  You have to work hard to find a 
> way to cause some of the common errors you see in C in other languages.

back when C was young, programmers were much more disciplined.  Probably
because every pre-pubescent dweeb without a girlfriend couldn't claim to
be a "programmer" and have the world accept it.  It is not the language
that causes the problem.  COBOL has no bounds checking.  Most Pascals will
let you turn off range and bounds checking.  Most Fortran compilers will
blindly let you define things in common blocks in different orders at
differnt points.  ADA, the language that was going to save us all.  I have
a text book on it here somewhere that devotes an entire chapter to how to
get around all the safeties like "strong typing" and "range and bounds
checking".  And the list just goes on and on.

Read my lips, it ain't the language!!!  And, it ain't the OS either.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

Bill Gunshannon wrote:

> De facto made it a standard.  To the best of my knowledge C does not have
> a string data type at all. 

char *mystring = "Hello world" ;

This defines a pointer to an area of data that contains the NULL
TERMINATED "Hello World" string.

So the compiler does in fact make use of null terminated strings.

In article <i8enuc$n5r$1@lnx107.hrz.tu-darmstadt.de>,
	m.kraemer@gsi.de (Michael Kraemer) writes:
> In article <7c5642b2-7d20-4d89-8a2e-b052f9fe02a8@26g2000yqv.googlegroups.com>, H
> Vlems <hvlems@freenet.de> writes:
>  
>> VMS (and DEC) has its influences in many areas of IT 
> 
> almost forgotten but probably most influential until now:
> X Window and Ethernet.
> 
>> but I never
>> considered DEC as a language development company.
> 
> it was a hardware company in the first place.
> 
>> BLISS is probably the best example, and it is about as appealing to me
>> as C is (I learned to program in Algol, in the Dijkstra school).
>> I don't think Ada would have made it as a system programming language,
>> but PL/1 might have done that. 
> 
> Way too heavy weight. Some platforms never saw a PL/I compiler,
> and the most popular ones only when it was much too late
> and C/C++ had filled the void.
> 
>> Pascal, like Algol60, is too simple.
>> It might have been extended, like Burroughs did with Algol, to a
>> language in which you could write operating systems and compilers.
> 
> Such a language has to be simple and readily available
> on whatever piece of silicon out there, which leaves - C.

Actually, because the compiler and support libs are readily written
in Pascal, Pascal makes an excelent bootstrap language.  One can
start out with only what is necessary do an OS (pretty much Byte
and Int) and then extend it as needed.  One usually only needs a
couple of routines written in assembler and they can frequently be
removed as development continues.  And all of this with a language
that was certainly not designed for it!!

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <Hv6dnba-gp22LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>,
	GreyCloud <mist@cumulus.com> writes:
> JF Mezei wrote:
>> smithfarm wrote:
>> 
>>> On the other hand, if VMS is ever to gain any traction as a viable "real 
>>> world" alternative to Windows and Linux, there is a certain minimal (and 
>>> shifting) feature set that it must support.
>> 
>> Rewind 10-12 years: Enters Steve Jobs to rescue Apple. He job was to
>> revamp the Macintosh product line and bring it back to life.  He used
>> "vision" to convince the board to spend the big bucks to redevelop the OS.
>> 
>> Bringing VMS back to life as a server OS would take less effort than it
>> took Apple to resurrect itself. Port it to the 64 bit 8086 and bring in
>> modern stuff like IPsec and true IPv6 support.
>> 
>> Porting the browser is in fact fairly smart because it also forced the
>> porting of much middleware such as php, perl and a buch of others that
>> are used to build the Mozilla product. But it must remain a continuous
>> effort to ensure that VMS has up-to-date middleware.
>> 
>> Also, if you need another OS to manage a server OS, you might as well
>> use that other OS as your server OS.
>> 
>> 
>> At this point in time we have to be realistic as painful as it may be.
>> HP has had 10 years to change its mind about VMS.  It hasn't. It won't.
>> 
>> With VMS management and engineering gone, there is nobody to fight for
>> VMS at HP.  And when the new guy inquires about HP's products, he will
>> probably be told that it is too late to revive VMS and that the
>> remaining VMS customers are happy with VMS being on maintenance mode
>> with limited new development. There will be nobody to counter this.
>> 
> 
> Why doesn't HP just put VMS up for sale?

Because they do not see any financial gain in doing so and money is their
most important product.

> 
>> Existing mamagement like Livermore will arrange for the new guy to meet
>> customers whose opinion of VMS matches what HP's policy on VMS is.
>> 
>> I don't know why anyone would still have hopes that HP would start a
>> major turn around for VMS.
> 
> It amazes me why HP bought VMS and turn around and deny its existence to 
> the world.  Do they get a tax write off or something?

HP did not buy VMS.  HP bought Compaq, a PC company.  Anything else that
came along on the deal is just excess baggage.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <1ucrn7-pib.ln1@nntp.local.net>,
	Captain Pugwash <alex.buell@munted.org.uk> writes:
> On Mon, 2010-10-04 at 04:18 -0400, JF Mezei wrote:
> 
>> I don't know why anyone would still have hopes that HP would start a
>> major turn around for VMS.=20
> 
> Can't someone just step up and buy the rights to OpenVMS from HP,

It's not for sale.

> become an independent vendor, and live off the sales and support
> revenue stream?

It is unlikely that this existing stream of revenue would return your
investment (if VMS was, in fact, for sale) in any period of time.
Certainly not before you filed for bankruptcy.

> 
> Then port like mad to other architectures such as x86 and AMD64?=20

Who is funding all this work?

> 
> Or even open source the whole thing.=20

Yeah, now there's a real return on your several million dollar (if
VMS was, in fact, for sale) investment.

Not much on business, are you?

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <m7hrn7-pib.ln1@nntp.local.net>,
	Captain Pugwash <alex.buell@munted.org.uk> writes:
> On Tue, 2010-10-05 at 19:53 +0800, Richard Maher wrote:
> 
>> There *is* enough money already!
>>
>> It's just that it's *still* being siphoned into the personal pet projects=
>  of
>> the scum that has survived 30 years without writing a single line of VMS =
> or
>> layered product code :-(
> 
> Just out of interest just how many organisations out there still use
> OpenVMS or VMS?
> 

OK, someone trot out The VMS Constant.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

Bill Gunshannon wrote:

> Certainly not before you filed for bankruptcy.

HP might do like oil companies do just before a well runs dry. They sell
it to a small company who gets to operate the well for a year or two,
and is then stuck with the cost of cleaning up the mess when the well is
decommissioned.

HP might sell VMS to some other outfit which would free HP from any
support obligations.

On Tue, 2010-10-05 at 18:14 +0000, Bill Gunshannon wrote:
>=20
> > Can't someone just step up and buy the rights to OpenVMS from HP,
>=20
> It's not for sale.

Why not?=20

> > become an independent vendor, and live off the sales and support
> > revenue stream?
>=20
> It is unlikely that this existing stream of revenue would return your
> investment (if VMS was, in fact, for sale) in any period of time.
> Certainly not before you filed for bankruptcy.

If that was true, HP would have sold it off by now.=20

> > Then port like mad to other architectures such as x86 and AMD64?=3D20
>=20
> Who is funding all this work?

Customers.=20

> > Or even open source the whole thing.=3D20
>=20
> Yeah, now there's a real return on your several million dollar (if
> VMS was, in fact, for sale) investment.

Only several million dollars? That's remarkably cheap. Microsoft <spit!>
Windows cost billions to develop. Linux would cost billions if all the
contributions to it was valued in terms of monetary value.=20

> Not much on business, are you?=20

We'll see.
--=20
Tactical Nuclear Kittens

In article <cxJeWhYXnsCo@eisner.encompasserve.org>,
koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:
> In article <i8fchd$sf$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de (Michael
> Kraemer) writes:
> > 
> > since nobody uses it outside of the VMS niche,
> > this is a bit hard to verify.
> 
>    Since BLISS was invented on a PDP-10,

gone.

>  ported to VAX, Alpha, and
>    Itanium, 

VMS niche, as said.

> targeted PDP-11 

gone.

> (via cross compilers), and is being
>    ported to IA-32,

in other words: not available, let alone being used.

> the claim that it is highly portable it's fairly easy 
>    to verify.

ported to mostly irrelevant platforms,
which leaves VMS, as I said.

>    If you don't know how different PDP-10 was from the rest of those,
>    shut up and go learn.

should I care, except for historical reasons? 

Bill Gunshannon <billg999@cs.uofs.edu> wrote:
> In article <i8dak6$5h9$1@usenet01.boi.hp.com>,
>        "FredK" <fred.nospam@dec.com> writes:
(snip)
 
>> Bill, I didn't mean to imply UNIX or C "invented" zwro terminated strings. 
 
> Not singling you out.  It is just a common thread here where people
> seem to ignore the fact that Unix/C merely picked up on a method that
> was pretty much standard accross the industry.

As far as I know, it was not commonly used by IBM at the time.
Pretty much the IBM code from OS/360 and successors uses count
fields with strings, even text strings.  (EBCDIC also has a NUL
at position X'00', so it could have been done.)
 
>> C made it the language standard, 
 
> De facto made it a standard.  To the best of my knowledge C 
> does not have a string data type at all. 
 
>>           UNIX enshrined it in it's interfaces. 

Isn't C older than unix?  In any case, Unix and C are closely
related, and seem to have helped propagate it.

Otherwise, it seems to me that DEC used it before C.
 
> You mean like the example I posted for VMS?  Use made it what it is and
> people are very unlikely to make major changes that break lots of already
> written software.  That is what caused all the two year dates more than a
> decade after punched-cards went away and keeps time_t too small and even
> keeps IPV6 out of the mainstream (I first used IPV6 more than 10 years
> ago and we actually had (have?) an address allocation).
 
> Hindsight is always 20/20.  It served a purpose and worked very well for
> a long time.  It is still usable and perfectly safe depending on the
> programmer.   And, in some environments it is perfectly safe regardless.
> I would foresee no problem using .ASCIZ cwunder RT-11, RSTS, RSX or under
> VMS.  Of course, ~i have no problem with using it under Unix, but that's
> me.  :-)

It doesn't work when you need to be able to send all the possible
bit patterns.

I do remember in the OS/360 days putting a carriage return in
a literal constant in Fortran.  

-- glen

In article <c7709661-68a9-4722-9106-bc3842706467@t20g2000yqa.googlegroups.com>, H Vlems <hvlems@freenet.de> writes:
> And even though BLISS
> is a highly protable product, it never
> took the world. 

   Wulf, Russell, and Habermann were at Carnegie Mellon University at
   the time they invented BLISS, so why would DEC get the credit?

   Just because programming in Macro-10 was such a PITA?

JF Mezei <jfmezei.spamnot@vaxination.ca> wrote:
> Bill Gunshannon wrote:
 
>> De facto made it a standard.  To the best of my knowledge C 
>> does not have a string data type at all. 
 
> char *mystring = "Hello world" ;
 
> This defines a pointer to an area of data that contains the NULL
> TERMINATED "Hello World" string.
 
> So the compiler does in fact make use of null terminated strings.

This is true, but the C language doesn't require that you use them.
You can write C code that doesn't use the str... library routines,
keeps track of string lengths somewhere else, and doesn't depend
on the null terminations supplied by the compiler.  If you really
want it,

#include <stdio.h>

int main() {
   char mystring[11]={"Hello world"};
   printf("%d\n",sizeof mystring);
   }

You can convince the compiler not to generate one.  (The string
may be padded, especially on word addressed machines.)

-- glen

In article <i8fchd$sf$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de (Michael Kraemer) writes:
> 
> since nobody uses it outside of the VMS niche,
> this is a bit hard to verify.

   Since BLISS was invented on a PDP-10, ported to VAX, Alpha, and
   Itanium, targeted PDP-11 (via cross compilers), and is being
   ported to IA-32, the claim that it is highly portable it's fairly easy 
   to verify.

   If you don't know how different PDP-10 was from the rest of those,
   shut up and go learn.

Bill Gunshannon <billg999@cs.uofs.edu> wrote:
(snip on malloc() and error return codes)

> Wrong.  The NULL is not an error return it is how much memory was allocated.

(someone else wrote) 
>> Returning NULL provides me with zero information that can be 
>> either used to correct the problem, or instruct the 
>> user how to correct the problem - other than guessing.
 
> Unless, you decide to read tha man page which tells you to check errno.
 
(snip) 
> There is only one possible failure of malloc().  No memory to allocate.

Well, some systems might provide for the reason that not enough
memory was available.  Maybe the fact that other users were using
it could be different from asking for more memory than existed.
Maybe fragmentation (enough memory, but not contiguous), 
or maybe a parity error detected while trying to allocate it.

> And it does return a separate error code even though there is only one.
> I can think of one possible reason why malloc() would return NULL and
> have it not be an error.  if you asked for NULL bytes.  Otherwise the
> only cause for a return of NULL would be not enough memory available to
> meet the request and that is handled perfectly well as documented.

It does seem common for malloc(0) to return NULL.  One possibility
is that it interpreted as asking for UINT_MAX+1 bytes, which 
likely fails (at least it used to).  In any case, it is commonly
known to be careful with malloc(0).

-- glen 

On 10/5/2010 1:56 PM, Bill Gunshannon wrote:
> In article<i8dq16$d9r$1@usenet01.boi.hp.com>,
> 	"FredK"<fred.nospam@dec.com>  writes:
>>
>> "John Wallace"<johnwallace4@yahoo.co.uk>  wrote in message
>> news:a7ed3d76-789f-49e6-b459-e6da503edcf2@a36g2000yqc.googlegroups.com...
>> On Oct 4, 11:13 pm, Bob Eager<rd...@spamcop.net>  wrote:
>>
>>>>>
>> Fred may not be 100% accurate in saying no error code is returned by
>> malloc(), but he is *right* about the Unix approach to error reporting
>> in general. Windows (with GetLastError()) is no better either.
>>>>>
>>
>> For some reason, my newsreader isn't quoting your response.
>>
>> Thanks.  Essentially because certain routines didn't include an error
>> return, they created the errno hack.  Quite a hack... a global symbol.
>
> Matter of opinion.  It has served Unix programmers well for a lot longer
> than VMS has even been around.
>
>>
>> I can defend things like C not having a general string type if C is viewed
>> as a assembler replacement as opposed to a high-level language (although a
>> counted string would make more sense in general than null terminated).
>
> If you want a counted string, create it.  That's one of the beauties of
> C.  The basic types as might be defined at the hardware level already
> exist and you are free to create any other user defined type you feel
> necessary.  Maker a string what you want a string to be.  I have worked
> with counted strings (ala UCSD Pascal) I saw no real advantage to them.
> And one very big disadvantage.  They come with a very finite fixed maximum
> length.  If nothing else, that either keeps them very short or makes them
> very non-portable.
>
>>
>> The language decision led to OS API decisions.
>
> How?  The language does not define a "string" type.  That is actually
> what is in the OS API.  And there have been numerous opportunities to
> change it.  No one showed any interest. (Or have we all forgotten things
> line "Safe C" which was created back in the 70's and as a commercial
> product, didn't fly.)
>
>>
>> I've been a programmer for a living since around 1977/78.
>
> I first programmed around 1971 but it was shortly after you that I took
> it up as a full-time job.
>
>>                                                              The bulk of the
>> code I've written in those 30+ years has been C.
>
> While I have 9and still do) write a lot of C I have actually written
> programs in well over a dozen different languages many of which I have
> long since forgotten.
>
>>                                                    I'd rather write C than
>> write BLISS.  But you have to be blind to ignore the programming abuse I've
>> seen in C and in the library interfaces.  You have to work hard to find a
>> way to cause some of the common errors you see in C in other languages.
>
> back when C was young, programmers were much more disciplined.  Probably
> because every pre-pubescent dweeb without a girlfriend couldn't claim to
> be a "programmer" and have the world accept it.  It is not the language
> that causes the problem.  COBOL has no bounds checking.  Most Pascals will
> let you turn off range and bounds checking.  Most Fortran compilers will
> blindly let you define things in common blocks in different orders at
> different points.  ADA, the language that was going to save us all.  I have
> a text book on it here somewhere that devotes an entire chapter to how to
> get around all the safeties like "strong typing" and "range and bounds
> checking".  And the list just goes on and on.
>
> Read my lips, it ain't the language!!!  And, it ain't the OS either.
>

The language and/or the OS can give idiots free rein!  Or not! 
Languages such as ADA try to minimize the scope of idiocy.  You can 
still write bad code but you have to work a lot harder to do it.

Idiots can be useful. One such managed to crash one of my systems by 
making a request to read zero bytes from a device.  I had to fix the 
driver to return an error when someone tried to read or write fewer than 
one byte.

He taught me something about defensive programming!

Bill Gunshannon <billg999@cs.uofs.edu> wrote:
(snip)
 
> Why?  The only cause of an error return from malloc(0 is if there isn't
> enough memory to meet the request.  ENOMEM is a numeric value intended
> to be interpreted and the programmer is free to return any human readable
> message to the user he wishes if he can't programatically fix the problem.
> ENOMEM is not intended for users.  If one actually looks at what WENOMEM
> means it is simply: "Can't allocate memory".  How much cleare can you be?
 
> And if you don't want to write your own message you merely call perror
> and it will take care of it for you. 

(snip)

Now, compare to the MacOS (before OS X) where a picture of a bomb
appears and ERROR -2 OCCURRED.

(I believe that -2 is right, though it has been a while.)

-- glen

In article <4caa5220$0$19446$c3e8da3$c8b7d2e6@news.astraweb.com>, JF
Mezei <jfmezei.spamnot@vaxination.ca> writes: 

> Porting VMS to the 8086 is not a magic bullet. Changing HP's mind on VMS
> is the magic bullet.

In the past, I was sceptical about porting VMS to the 8086 (we all know 
what JF means by that).  At the time, I think I was right.  However, 
other hardware is now better than Itanium, it is unclear if Itanium has 
a viable future and the technical difficulties of going to the 8086 are 
probably not much worse than going to Itanium.  Now, VMS is tied to a 
chip.  If that chip is not viable, then VMS is in trouble.  Yes, 
customers can and do pay for the special chip in addition to VMS.  
However, a) there are those who need faster chips, and VMS no longer 
runs on the fastest hardware and b) paying a premium for a special chip 
which is SLOWER than the competition is more difficult.

In article <1ucrn7-pib.ln1@nntp.local.net>, Captain Pugwash
<alex.buell@munted.org.uk> writes: 

> Can't someone just step up and buy the rights to OpenVMS from HP,
> become an independent vendor, and live off the sales and support
> revenue stream?

Do the math.

In article <p-qdnSOQBuqTHjbRnZ2dnUVZ_vydnZ2d@giganews.com>,
	"Richard B. Gilbert" <rgilbert88@comcast.net> writes:
> On 10/5/2010 1:56 PM, Bill Gunshannon wrote:
>> In article<i8dq16$d9r$1@usenet01.boi.hp.com>,
>> 	"FredK"<fred.nospam@dec.com>  writes:
>>>
>>> "John Wallace"<johnwallace4@yahoo.co.uk>  wrote in message
>>> news:a7ed3d76-789f-49e6-b459-e6da503edcf2@a36g2000yqc.googlegroups.com...
>>> On Oct 4, 11:13 pm, Bob Eager<rd...@spamcop.net>  wrote:
>>>
>>>>>>
>>> Fred may not be 100% accurate in saying no error code is returned by
>>> malloc(), but he is *right* about the Unix approach to error reporting
>>> in general. Windows (with GetLastError()) is no better either.
>>>>>>
>>>
>>> For some reason, my newsreader isn't quoting your response.
>>>
>>> Thanks.  Essentially because certain routines didn't include an error
>>> return, they created the errno hack.  Quite a hack... a global symbol.
>>
>> Matter of opinion.  It has served Unix programmers well for a lot longer
>> than VMS has even been around.
>>
>>>
>>> I can defend things like C not having a general string type if C is viewed
>>> as a assembler replacement as opposed to a high-level language (although a
>>> counted string would make more sense in general than null terminated).
>>
>> If you want a counted string, create it.  That's one of the beauties of
>> C.  The basic types as might be defined at the hardware level already
>> exist and you are free to create any other user defined type you feel
>> necessary.  Maker a string what you want a string to be.  I have worked
>> with counted strings (ala UCSD Pascal) I saw no real advantage to them.
>> And one very big disadvantage.  They come with a very finite fixed maximum
>> length.  If nothing else, that either keeps them very short or makes them
>> very non-portable.
>>
>>>
>>> The language decision led to OS API decisions.
>>
>> How?  The language does not define a "string" type.  That is actually
>> what is in the OS API.  And there have been numerous opportunities to
>> change it.  No one showed any interest. (Or have we all forgotten things
>> line "Safe C" which was created back in the 70's and as a commercial
>> product, didn't fly.)
>>
>>>
>>> I've been a programmer for a living since around 1977/78.
>>
>> I first programmed around 1971 but it was shortly after you that I took
>> it up as a full-time job.
>>
>>>                                                              The bulk of the
>>> code I've written in those 30+ years has been C.
>>
>> While I have 9and still do) write a lot of C I have actually written
>> programs in well over a dozen different languages many of which I have
>> long since forgotten.
>>
>>>                                                    I'd rather write C than
>>> write BLISS.  But you have to be blind to ignore the programming abuse I've
>>> seen in C and in the library interfaces.  You have to work hard to find a
>>> way to cause some of the common errors you see in C in other languages.
>>
>> back when C was young, programmers were much more disciplined.  Probably
>> because every pre-pubescent dweeb without a girlfriend couldn't claim to
>> be a "programmer" and have the world accept it.  It is not the language
>> that causes the problem.  COBOL has no bounds checking.  Most Pascals will
>> let you turn off range and bounds checking.  Most Fortran compilers will
>> blindly let you define things in common blocks in different orders at
>> different points.  ADA, the language that was going to save us all.  I have
>> a text book on it here somewhere that devotes an entire chapter to how to
>> get around all the safeties like "strong typing" and "range and bounds
>> checking".  And the list just goes on and on.
>>
>> Read my lips, it ain't the language!!!  And, it ain't the OS either.
>>
> 
> The language and/or the OS can give idiots free rein!  Or not! 
> Languages such as ADA try to minimize the scope of idiocy.  You can 
> still write bad code but you have to work a lot harder to do it.

Interestingly enough, idiots don't seem to have to work very hard to
remain idiots.

> 
> Idiots can be useful. One such managed to crash one of my systems by 
> making a request to read zero bytes from a device.  I had to fix the 
> driver to return an error when someone tried to read or write fewer than 
> one byte.
> 
> He taught me something about defensive programming!
 
And how many "programmers" have learned absolutely nothing from all the
problems caused by bad programs?  It is always much easier to just blame
the language.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8g040$rqj$2@speranza.aioe.org>,
	glen herrmannsfeldt <gah@ugcs.caltech.edu> writes:
> Bill Gunshannon <billg999@cs.uofs.edu> wrote:
> (snip)
>  
>> Why?  The only cause of an error return from malloc(0 is if there isn't
>> enough memory to meet the request.  ENOMEM is a numeric value intended
>> to be interpreted and the programmer is free to return any human readable
>> message to the user he wishes if he can't programatically fix the problem.
>> ENOMEM is not intended for users.  If one actually looks at what WENOMEM
>> means it is simply: "Can't allocate memory".  How much cleare can you be?
>  
>> And if you don't want to write your own message you merely call perror
>> and it will take care of it for you. 
> 
> (snip)
> 
> Now, compare to the MacOS (before OS X) where a picture of a bomb
> appears and ERROR -2 OCCURRED.
> 
> (I believe that -2 is right, though it has been a while.)
> 

What's your point?  I have always considered MacOS to be a bunch of crap.
I still have a whole bunch of M68K Macs.  No one was able to get less
performance out of a Motorolla CPU than Apple.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <4cab6730$0$30024$c3e8da3$76a7c58f@news.astraweb.com>, JF Mezei <jfmezei.spamnot@vaxination.ca> writes:
> Bill Gunshannon wrote:
> 
>> De facto made it a standard.  To the best of my knowledge C does not have
>> a string data type at all. 
> 
> char *mystring = "Hello world" ;
> 
> This defines a pointer to an area of data that contains the NULL
> TERMINATED "Hello World" string.
> 
> So the compiler does in fact make use of null terminated strings.

   C does not have a string variable type, but it does have string
   constants.  So it's a semantical split to determine weather C has
   a string "data type".  Reminds me of Fortran IV.

   Imagine a programming language with integer constants but no
   integer variables, and you can see how absurd this is.

In article <i8fsc7$83q$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de (Michael Kraemer) writes:
> 
> gone.

   That and your toher comments ignore that "ported" _is_ past tense.
   And is sufficient to measure portability.

   How would you measure it, by the number of platforms that support
   D tomorrow?

In article <i8fv0c$q0c$1@speranza.aioe.org>, glen herrmannsfeldt <gah@ugcs.caltech.edu> writes:
> 
> Isn't C older than unix?  In any case, Unix and C are closely
> related, and seem to have helped propagate it.

   Invented at about the same time.  UNIX was not originally written in C.  
   At the time of one of its rewrites C, BLISS, and other languages
   were considered.

Jan-Erik Soderholm wrote:
> On 2010-10-05 07:04, GreyCloud wrote:
> 
>> But as I asked JF, why doesn't HP just sell VMS to someone else if they
>> continue to hide it from the world on their web site?
> 
> Do they ?
> 
> If I go to www.hp.com (and beeing redirectec to the swedish HP site),
> I can look under "Servers" and have OpenVMS listed together
> with the other OS'es. Or click "Software" and have OpenVMS listed
> on the next page again together with all the other OS'es.
> 
> In what way is OpenVMS hided from the world any more then any of
> the other operating systems ?
> 

I have to click on Large Businesses and then click on servers.
And there is one lone OpenVMS entry in a list.  Not on the front page, 
that much is certain.  Their front page is basically windows oriented 
with no mention of OpenVMS at all, let alone HP-UX.  I consider this 
poor marketing of all of their products.  Worse, if you want a price on 
any of their high end products, you have to invite a call from a 
salesman... something I absolutely abhor.

If I go to M$ website, obviously you'll see their operating system 
advertised along with all their flagship products.

Michael Kraemer wrote:
> In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>, GreyCloud
> <mist@cumulus.com> writes:
> 
>> But as I asked JF, why doesn't HP just sell VMS to someone else 
> 
> Why should they?
> They don't have a problem with it.
> Contrary to common belief here the world doesn't have problems
> without it.
> So why change the status quo?

I thought HP as a company was in business to sell products.
Either advertise it or sell it off.

> 
>> if they 
>> continue to hide it from the world on their web site? 
> 
> Do they?
> So far you would find all VMS related info on their respective pages.
> Of course not on the front pages, no wonder
> since 99+x% of HP customers are interested in PCs and ink
> rather than nerd stuff such as VMS. 
> 

Believe me, it isn't out in the open at HPs website... but their PCs 
being MS OS types are right there in your face at their front page.

It would be like IBM not advertising their OS... which they do and the 
world knows it.

Bill Gunshannon wrote:
> In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>,
> 	GreyCloud <mist@cumulus.com> writes:
>> Michael Kraemer wrote:
>>> In article <4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com>, JF Mezei
>>> <jfmezei.spamnot@vaxination.ca> writes:
>>>> Also, if you need another OS to manage a server OS, you might as well
>>>> use that other OS as your server OS.
>>>>
>>> exactly.
>>> Moreover, to come up with another of those popular
>>> car analogies: what some people suggest here is
>>> to sell a car restricted to engine and tank ("server-centric").
>>> If you want to actually drive it,
>>> you'd have to buy/use a second one,
>>> which will give you seats, tires and a driving wheel,
>>> so you can hook up the first one.
>>> I think most, if not all customers would just
>>> buy the second, complete one, even if it 
>>> has a "weaker" (by whatever measure) engine
>>> than the first one.
>> You've got a good point.
>> Look at cummins diesel plant.  They don't make complete trucks, but they 
>> do sell their engines to other companies that do.  One would thing that 
>> HP could sell their engine to others that would complete the product 
>> offering.
>>
>> But as I asked JF, why doesn't HP just sell VMS to someone else if they 
>> continue to hide it from the world on their web site?  Something just 
>> doesn't add up here.
> 
> Probably because the perciened cost of selling the product line is greater
> than any percieved benefit.
> 

That would really depend on what company would want to buy it and how 
well they can enhance the product to be something desirable for the 
masses.  After all, it took someone like Steve Jobs to take a unix os 
and transform it into what it is now.  I see no reason why the same 
can't be done with VMS.

Bob Koehler wrote:
> In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>, GreyCloud <mist@cumulus.com> writes:
>> But as I asked JF, why doesn't HP just sell VMS to someone else if they 
>> continue to hide it from the world on their web site?  Something just 
>> doesn't add up here.
> 
>    Why sell a product to a potential competitor that could then
>    actually compete with you?
> 
>    There could be some way to sell lots of ink if you owned VMS.  8-)
> 

I can see that angle.  :-))

Still tho... HP hides VMS very well.

GreyCloud wrote:
> Bob Koehler wrote:

> Still tho... HP hides VMS very well.


HP doesn't actively hide VMS. They just don't promote it. It is there,
itn its little corner, visible to those who look for it, but invisible
to those passing by.

No amount of bickering will bring VMS back. If I have accepted it and
moved on, then everyone else should.

Those lucky enough to still have gainful VMS related employment should
cherish it.

On 5 okt, 16:25, m.krae...@gsi.de (Michael Kraemer) wrote:
> In article <c7709661-68a9-4722-9106-bc3842706...@t20g2000yqa.googlegroups.com>,
>
> H Vlems <hvl...@freenet.de> writes:
>
> > It's not easy to summarize what DEC achieved in a couple of words.
> > First and foremost a hardware company
> > but DEC also built fine software products.Their OSes were DEC's own
> > design. DEC also had an excellent
> > reputation for building compilers,
>
> Not for me. They had just as many bugs
> as the competitor's, if not more,
> in particular on DEC's RISC machines.
> Could be, of course, that they just weren't given
> enough time to mature, unlike on VAX.

I'm not familiar with the DEC RISC systems. I have one of those and it
was used as a disk
expansion cabinet for a VAXstation. IIRC a lot of their systems
software came from a unix
flavour. The DEC supplied products on VMS and RT-11 never gave me
problems.
>
> > its members continue to do so but
> > now for Intel and others.
> > And no, I didn't forget about ethernet. The point was whether DEC did
> > design a programming language that
> > had as much influence as its other products did. And even though BLISS
> > is a highly protable product,
>
> since nobody uses it outside of the VMS niche,
> this is a bit hard to verify.
>
Which means BLISS for the VAX, the AXP and the IA64. And it was
available for the PDP-10 and
DECsystem 20 series, on the PDP-11 for various operating systems.
DEC did a lot of research on compiler development, like splitting the
final code separation and the
language dependent parsing stages so that these stages could be used
on other compilers or platforms.
ISTR that theconcept for  elf compilers came from DEC too.

> > it never
> > took the world. Neither did Burroughs Extended Algol. Portability
> > isn't the only issue otherwise linux would have
> > been written in COBOL.
>
> I can't see why Cobol should be more portable than C.

Very strict language syntax and the possibility for tight control on
variable declarations.
In most languages integer i gets you a platform dependent integer
range. In COBOL you
can specify your needs.

> > PL/I is a rich language with lots of functionality to write system
> > software in.
>
> my experience from a former life is just the opposite.
> It was good for high level stuff like formatted I/O,
> but as soon as things came closer to OS and hardware,
> assembly language was a much better choice.

Assembly language is an obvious choice but there are systems out there
without an assembler.
PL/M was an attempt to make do without assembly language for small
microprocessor based
(embedded) systems. There used to be a cross compiler on VMS
available.

> > It may be too rich, that's probably why
> > subsets came into existence for exactly that purpose.
>
> and there goes portability.
> If supported subsets differ from one compiler
> to another, you loose.
>
> > There are quite
> > a lot of proprietary derivatives, which probably
> > is the reason why they never took off.
>
> Exactly. There are too many constructs defined in the
> language itself, making it impossible to even compile
> certain PL/I sources on a different platform.
> For example, unless the standard has evolved significantly
> in the past 20 years, stuff like
>
> DCL FILE ENV(vsam);
>
Platform dependent constructs and language extensions are as portable
as the platform.
That is true for any language.

> won't compile properly on anything not supporting that IBM stuff.
> On C one would use sth like
>
> fp = fopen( filename, "type=vsam" );
>
> which would at least compile and one can hide
> platform dependent stuff in literals and/or ifdef's.
>
> > And no, C is not always easy to implement, ask Unisys.
>
> who's Unisys?
> And, if they aren't even able to implement a language
> as simple as C, should I want to know? :-)

Unisys = Sperry + Burroughs.
Burroughs started development on pure stack operated machines in the
late 50-ies.
They decided that assembly language wouldn't do for systems
programming on a modern system.
They handcoded an (Extended Algol) compiler which was used for the
next version of the Algol
compiler and the operating system. Google for Waychoff for anecdotes
on this topic.

Hans

On 5 okt, 21:41, glen herrmannsfeldt <g...@ugcs.caltech.edu> wrote:
> Bill Gunshannon <billg...@cs.uofs.edu> wrote:
> > In article <i8dak6$5h...@usenet01.boi.hp.com>,
> > =A0 =A0 =A0 =A0"FredK" <fred.nos...@dec.com> writes:
>
> (snip)
>
> >> Bill, I didn't mean to imply UNIX or C "invented" zwro terminated stri=
ngs.
> > Not singling you out. =A0It is just a common thread here where people
> > seem to ignore the fact that Unix/C merely picked up on a method that
> > was pretty much standard accross the industry.
>
> As far as I know, it was not commonly used by IBM at the time.
> Pretty much the IBM code from OS/360 and successors uses count
> fields with strings, even text strings. =A0(EBCDIC also has a NUL
> at position X'00', so it could have been done.)
>
> >> C made it the language standard,
> > De facto made it a standard. =A0To the best of my knowledge C
> > does not have a string data type at all.
> >> =A0 =A0 =A0 =A0 =A0 UNIX enshrined it in it's interfaces.
>
> Isn't C older than unix? =A0In any case, Unix and C are closely
> related, and seem to have helped propagate it.
>
> Otherwise, it seems to me that DEC used it before C.
>
> > You mean like the example I posted for VMS? =A0Use made it what it is a=
nd
> > people are very unlikely to make major changes that break lots of alrea=
dy
> > written software. =A0That is what caused all the two year dates more th=
an a
> > decade after punched-cards went away and keeps time_t too small and eve=
n
> > keeps IPV6 out of the mainstream (I first used IPV6 more than 10 years
> > ago and we actually had (have?) an address allocation).
> > Hindsight is always 20/20. =A0It served a purpose and worked very well =
for
> > a long time. =A0It is still usable and perfectly safe depending on the
> > programmer. =A0 And, in some environments it is perfectly safe regardle=
ss.
> > I would foresee no problem using .ASCIZ cwunder RT-11, RSTS, RSX or und=
er
> > VMS. =A0Of course, ~i have no problem with using it under Unix, but tha=
t's
> > me. =A0:-)
>
> It doesn't work when you need to be able to send all the possible
> bit patterns.
>
> I do remember in the OS/360 days putting a carriage return in
> a literal constant in Fortran. =A0
>
> -- glen

No, unix is definitely older than C. When unix was developed on the
first platform, a bare PDP-7,
they used assembly language to build the os. The next version was
rewritten in a language that
derived from BCPL (a British invention). The port to the PDP-11 saw an
upgrade to this and that
became known as C.
Hans

In article <S_WdnR17l-oKljHRnZ2dnUVZ_tqdnZ2d@bresnan.com>, GreyCloud
<mist@cumulus.com> writes:

> 
> I have to click on Large Businesses and then click on servers.
> And there is one lone OpenVMS entry in a list.  Not on the front page, 
> that much is certain.  Their front page is basically windows oriented 

as are probably 95+x% of their customers.
Call it "customer oriented".

> with no mention of OpenVMS at all, let alone HP-UX.  I consider this 
> poor marketing of all of their products.  Worse, if you want a price on 
> any of their high end products, you have to invite a call from a 
> salesman... something I absolutely abhor.
> 
> If I go to M$ website, obviously you'll see their operating system 
> advertised along with all their flagship products.

So does HP. 
Their flagship products are PCs, printers, ink.
HP-UX etc are just figleafs to stay in enterprise business.

In article <S_WdnRx7l-qikTHRnZ2dnUVZ_tqdnZ2d@bresnan.com>, GreyCloud
<mist@cumulus.com> writes:

> 
> I thought HP as a company was in business to sell products.

But they do.

> Either advertise it or sell it off.

Probably they found out that the regular business crowd
isn't susceptible to PC-style advertising.

 
> Believe me, it isn't out in the open at HPs website... but their PCs 
> being MS OS types are right there in your face at their front page.

Because these are their main products.
I suggest you study one of HP's recent business reports
to learn about the (un)importance of VMS in relation to commodity products.

> It would be like IBM not advertising their OS... which they do and the 
> world knows it.

If you look at their front page you couldn't tell they
are selling computers or OS's at all. Just that they
have something to do with IT in general
(which is intentional, I guess).
You have to dig into their product pages to learn more.

In article <bd35ea7a-2190-46fb-8918-a20ab5868949@l20g2000yqm.googlegroups.com>,
H Vlems <hvlems@freenet.de> writes:
 
> No, unix is definitely older than C. When unix was developed on the
> first platform, a bare PDP-7,
> they used assembly language to build the os. The next version was
> rewritten in a language that
> derived from BCPL 

as are parts of the AmigaOS ...

> (a British invention). 

yes, the equivalent to driving on the wrong side of the road :-)

In article <93ba7a91-874e-42a0-88d0-0ef324812230@n26g2000yqh.googlegroups.com>,
H Vlems <hvlems@freenet.de> writes:

> 
> I'm not familiar with the DEC RISC systems. 

But sure you know what an Alpha is?
The first round of compilers (Fortran,C at least)
on both, VMS and Tru64 had their own share
of bugs, even years after their introduction.
Not any different from other vendor's 
version 1.0 products.

> I have one of those and it
> was used as a disk
> expansion cabinet for a VAXstation. 

I don't know any RISC box which could be
used as an expansion cabinet.
However, the DECstation 5000/2xx series
had no internal storage bays and thus needed
an external "storage expansion" box with the same
form factor,
which could of course be hooked up to any
SCSI host adapter.

> IIRC a lot of their systems
> software came from a unix
> flavour. The DEC supplied products on VMS and RT-11 never gave me
> problems.

Either because you're accustomed to the problems or
because the products were given enough time to mature.
By the time such comparisons could have been made,
say 1992/93,
the VAX products were close to 15 years,
the Mips products just three years
(as were most RISC competitor's) 
and Alpha products were barely beyond beta release.
The product quality was more a function of 
"time-on-market" rather than parent company. 


> >
> Which means BLISS for the VAX, the AXP and the IA64. And it was
> available for the PDP-10 and
> DECsystem 20 series, on the PDP-11 for various operating systems.

all of which are rather irrelevant these days.
Do we have that language on more relevant platforms,
say AIX, HP-UX, Linux, Solaris, Windows, zOS
(in alphabetical order)?
Do people actually use it for relevant products on those platforms?

> 
> Very strict language syntax and the possibility for tight control on
> variable declarations.
> In most languages integer i gets you a platform dependent integer
> range. In COBOL you
> can specify your needs.

This doesn't make it more portable, just more heavy weight,
just as PL/I.

> Platform dependent constructs and language extensions are as portable 
> as the platform.
> That is true for any language.

PL/I is full of them, and of course people will use them.
In contrast, it is rather hard to code platform dependent
language constructs in C.

Michael Kraemer wrote:

> So does HP. 
> Their flagship products are PCs, printers, ink.
> HP-UX etc are just figleafs to stay in enterprise business.

I would rephrase it:

Their flagship products are ink, industry standard servers, printers and
PCs.

(listed in order of profit margins).

I think that the real status of HP-UX/VMS/NSK within HP is to be
determined.  I get the feeling that they are being kept alive in order
to keep customers paying support money and keep IA64 sales going.

This may or may not change with the new CEO who wants greater focus on
software.

In article <S_WdnRx7l-qikTHRnZ2dnUVZ_tqdnZ2d@bresnan.com>,
	GreyCloud <mist@cumulus.com> writes:
> Michael Kraemer wrote:
>> In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>, GreyCloud
>> <mist@cumulus.com> writes:
>> 
>>> But as I asked JF, why doesn't HP just sell VMS to someone else 
>> 
>> Why should they?
>> They don't have a problem with it.
>> Contrary to common belief here the world doesn't have problems
>> without it.
>> So why change the status quo?
> 
> I thought HP as a company was in business to sell products.
> Either advertise it or sell it off.

I will say it again, but I really don't think anyone here wants to
hear it.

As a facet of business, it is possible that the cost of selling the
VMS would be greater than any expected return.  In that case, why
would they sell it?  They can let it sit there until it just goes
away collecting what little revenue it generates while putting the
bare minimum into it (can you say "get rid of all those high priced
engineers to reduce those costs).  And, all this still hinges on
there being a buyer who is willing to pay what they would expect to
get for it, which is also very unlikely.  This isn't like the local
fleamarket.  You don't throw it on the table and take whatever some-
one comes along and offers.

ooobill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <S_WdnR97l-pXkTHRnZ2dnUVZ_tqdnZ2d@bresnan.com>,
	GreyCloud <mist@cumulus.com> writes:
> Bill Gunshannon wrote:
>> In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>,
>> 	GreyCloud <mist@cumulus.com> writes:
>>> Michael Kraemer wrote:
>>>> In article <4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com>, JF Mezei
>>>> <jfmezei.spamnot@vaxination.ca> writes:
>>>>> Also, if you need another OS to manage a server OS, you might as well
>>>>> use that other OS as your server OS.
>>>>>
>>>> exactly.
>>>> Moreover, to come up with another of those popular
>>>> car analogies: what some people suggest here is
>>>> to sell a car restricted to engine and tank ("server-centric").
>>>> If you want to actually drive it,
>>>> you'd have to buy/use a second one,
>>>> which will give you seats, tires and a driving wheel,
>>>> so you can hook up the first one.
>>>> I think most, if not all customers would just
>>>> buy the second, complete one, even if it 
>>>> has a "weaker" (by whatever measure) engine
>>>> than the first one.
>>> You've got a good point.
>>> Look at cummins diesel plant.  They don't make complete trucks, but they 
>>> do sell their engines to other companies that do.  One would thing that 
>>> HP could sell their engine to others that would complete the product 
>>> offering.
>>>
>>> But as I asked JF, why doesn't HP just sell VMS to someone else if they 
>>> continue to hide it from the world on their web site?  Something just 
>>> doesn't add up here.
>> 
>> Probably because the perciened cost of selling the product line is greater
>> than any percieved benefit.
>> 
> 
> That would really depend on what company would want to buy it and how 
> well they can enhance the product to be something desirable for the 
> masses.  

That has nothing to do with the cost of selling it.  It's not a bargain-
baement PC.  Sale of VMS would involve lawyers and contracts and god
only knows what else.  It could easily (and probably would) cost more 
to transfer VSM than they could hope to find a sucker to pay.

>           After all, it took someone like Steve Jobs to take a unix os 
> and transform it into what it is now.  

Oh please.  Apple still makes up a very small part of the true Unix
market (remember most of the users of things like iPhones and iPads
have no idea and couldn't care less that it runs a version of Unix)
and their love is for Aapple and not Unix.  It's religion, not tech-
nology.  And then we have Linux which has considerably more market
share thanApple will ever see.  Jobs is just very good at selling
snake-oil.  I saw all the presentations fromt he realease of the new
iPhone and it reminded very strongly of a traveling medicine show.

>                                        I see no reason why the same 
> can't be done with VMS.

"It's the applications  stupid!" (Not attacking, just paraphrasing something
most people will probably remember. :-)

The amount of time, effort and money it would take it would take to
catch up now pretty much leaves VMS out of the mainstream for the
rest of its existence, however long that may be.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8habm$smb$1@lnx107.hrz.tu-darmstadt.de>,
	m.kraemer@gsi.de (Michael Kraemer) writes:
> In article <93ba7a91-874e-42a0-88d0-0ef324812230@n26g2000yqh.googlegroups.com>,
> H Vlems <hvlems@freenet.de> writes:
> 
>> Which means BLISS for the VAX, the AXP and the IA64. And it was
>> available for the PDP-10 and
>> DECsystem 20 series, on the PDP-11 for various operating systems.
> 
> all of which are rather irrelevant these days.
> Do we have that language on more relevant platforms,
> say AIX, HP-UX, Linux, Solaris, Windows, zOS
> (in alphabetical order)?
> Do people actually use it for relevant products on those platforms?
> 

Which brings up another interesting question!!!

Given that people have found it not prohibitively difficult to either
graft front-ends onto the Gun Compiler or write translators to C (often
as a precursor to a later GCC Ffrontend) has anyone ever given any
consideration to makeing a freely available Bliss or DIBOL compiler?

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <S_WdnR97l-pXkTHRnZ2dnUVZ_tqdnZ2d@bresnan.com>, GreyCloud
<mist@cumulus.com> writes:

> After all, it took someone like Steve Jobs to take a unix os 
> and transform it into what it is now.  I see no reason why the same 
> can't be done with VMS.

He didn't "transform a unix os".
He essentially trashed the old OS 9, took a BSD Unix as is,
stacked a new GUI and a couple of cool apps on top of it,
and sold it as OS X. Together with the then new G4 and G5 PPCs
that was an attractive product sold in numbers of 1 ore 2M per year,
so the costs could be recouped quick enough.
There's zero chance to repeat that for VMS because
all ingredients are missing.
Certainly you wouldn't like the old VMS be trashed and replaced
by some Unix (but maybe I'm wrong, given the recent comments on
VMS 8.4 :-). A new GUI and cool desktop apps are forbidden 
by the authorities here in c.o.v because VMS "is as server OS",
so out the door with the prospect of selling 1 or 2M units per year
to recoup costs before entering bankrupcy.
And, finally, there's no Itanic equivalent in sight for the G4 and
G5 CPUs. 

In article <8h3c89Fgc1U4@mid.individual.net>, billg999@cs.uofs.edu (Bill
Gunshannon) writes:
> 
> Which brings up another interesting question!!!
> 
> Given that people have found it not prohibitively difficult to either
> graft front-ends onto the Gun Compiler or write translators to C (often
                            ^^^^^^^^^^^^
hmmm, I"m wondering what that could be ...
a new secret weapon?

> as a precursor to a later GCC Ffrontend) has anyone ever given any
> consideration to makeing a freely available Bliss or DIBOL compiler?

It probably will end up stuck like "pligcc",
an attempt to create a free PL/I compiler.

> The amount of time, effort and money it would take it would take to
> catch up now pretty much leaves VMS out of the mainstream for the
> rest of its existence, however long that may be.

Stop it! You're disrupting my dreams of a return to the glory days.

On 10/06/2010 09:35 PM, Bill Gunshannon wrote:
> Given that people have found it not prohibitively difficult to either
> graft front-ends onto the Gun Compiler or write translators to C (often
> as a precursor to a later GCC Ffrontend) has anyone ever given any
> consideration to makeing a freely available Bliss or DIBOL compiler?
>

Yes to the BLISS compiler (don't really know anything about DIBOL).
I work on it when I get time (which is not often at the moment).  Maybe
it might even see some kind of release in the next 18 months.  That is
an optimistic target though.

Tim.

In article <i8hurl$5pq$1@lnx107.hrz.tu-darmstadt.de>,
	m.kraemer@gsi.de (Michael Kraemer) writes:
> In article <8h3c89Fgc1U4@mid.individual.net>, billg999@cs.uofs.edu (Bill
> Gunshannon) writes:
>> 
>> Which brings up another interesting question!!!
>> 
>> Given that people have found it not prohibitively difficult to either
>> graft front-ends onto the Gun Compiler or write translators to C (often
>                             ^^^^^^^^^^^^
> hmmm, I"m wondering what that could be ...
> a new secret weapon?

Ssorry, dyslexic fingers.  but I didn't post a correction as I thought
people wold figure it out when they saw GCC mentioned in the next line.
> 
>> as a precursor to a later GCC Ffrontend) has anyone ever given any
>> consideration to makeing a freely available Bliss or DIBOL compiler?
> 
> It probably will end up stuck like "pligcc",
> an attempt to create a free PL/I compiler.

Well, that was the case with COBOL at one time (I remember early requests
being met with the comment "you can't convert COBOL into C, it just won't
work") but eventually, it was done and now there are at least three that
I know of.

DIBOL would be fun, but Bliss might be really useful.  Another one of
my "looking at the future" retirement projects is to re-write some
version of Unix in a (supposedly) safer language with ADA being my
top runner right now.  It would be interesting from the standpoint of
not just the conversion but also benchmarking it to see what loss or
gain of efficiency might result.  Doing it in Bliss would be equally
interesting and I expect that even an early Bliss compiler would give
more efficient code than ADA.

bill
 

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <poKUlBa6k0b9@eisner.encompasserve.org>,
	koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:
>  On 10/06/2010 09:35 PM, Bill Gunshannon wrote:
>> Given that people have found it not prohibitively difficult to either
>> graft front-ends onto the Gun Compiler or write translators to C (often
>> as a precursor to a later GCC Ffrontend) has anyone ever given any
>> consideration to makeing a freely available Bliss or DIBOL compiler?
> 
>    The BLISS compiler for VMS has been freely available since soon
>    after VMS was ported to Alpha.
> 
>    There is a free BLISS compiler being developed in relationship
>    too FreeVMS.  You can get early releases, including the source.

If you read what I wrote, you would have seen that I am interested in
writting Bliss porgrams for VMS.  I already know that works.  I am
interested in seeing if it is possible to create a version of Unix
that lacks even the few small security problems that currently may
exist as a result of it being written in C. (Not saying there are any
and there really are none I know of that haven't been addressed in
the C version.  Just have never garnered any interest among the unix
community.)
> 
>    There is a source for a PDP-11 BLISS compiler freely available,  
>    IIRC it's a cross compiler meant to run on VMS.

Now that I would be interested in.  The only PDP-11 Bliss compiler
I was aware of was for the PDP-10 and was not available in source
anywhere I could find it.  Given a Bliss compiler in source to start
with I might try writting a Bliss to C translator myself (in my
copious freetime. :-)  Unless, of course it is like the current state
of "free" ADA.

bill



-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8huiu$5ob$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de (Michael Kraemer) writes:
> A new GUI and cool desktop apps are forbidden 
> by the authorities here in c.o.v because VMS "is as server OS"

   Says who?  Just because HP says it's a server OS, and some folks
   connected to HP follow c.o.v does not make that the opinion
   of folks on c.o.v.  And we'll take our own view of who constitutes
   authorities.

   Of course, it would be authorities in HP, not in c.o.v, that
   determine whether cool apps ship with VMS.  But the rest of
   us won't wait for HP to supply all our apps.
   

 On 10/06/2010 09:35 PM, Bill Gunshannon wrote:
> Given that people have found it not prohibitively difficult to either
> graft front-ends onto the Gun Compiler or write translators to C (often
> as a precursor to a later GCC Ffrontend) has anyone ever given any
> consideration to makeing a freely available Bliss or DIBOL compiler?

   The BLISS compiler for VMS has been freely available since soon
   after VMS was ported to Alpha.

   There is a free BLISS compiler being developed in relationship
   too FreeVMS.  You can get early releases, including the source.

   There is a source for a PDP-11 BLISS compiler freely available,  
   IIRC it's a cross compiler meant to run on VMS.

Bill Gunshannon wrote:

> Oh please.  Apple still makes up a very small part of the true Unix
> market 

and later on:


> "It's the applications  stupid!" (Not attacking, just paraphrasing something
> most people will probably remember. :-)


You should know that OS-X is more Unix than you think.

With support for X11, GTK etc, OS-X is in fact just as Unixy a the guy
next door. The Apple servers come with standard Unix applications
(bind9, postfix, dovecot, jabber, ftp, etc etc). And you can "make" just
about any application designed for unix on a Mac.


So, in the "its the applications stupid!" department, OS-X is very much
Unix and, when you combine that with the OS-X "native" applications that
use the Apple proprietary frameworks and COCOA UI, you get a fairly
impressive application portfolio.

Windows does not come out of the box with unix compatibility.

koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:

> In article <i8fchd$sf$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de
> (Michael Kraemer) writes:

>> since nobody uses it outside of the VMS niche,
>> this is a bit hard to verify.

>    Since BLISS was invented on a PDP-10, ported to VAX, Alpha, and
>    Itanium, targeted PDP-11 (via cross compilers), and is being
>    ported to IA-32, the claim that it is highly portable it's fairly easy 
>    to verify.

>    If you don't know how different PDP-10 was from the rest of those,
>    shut up and go learn.

In point of fact, the later Fortran compiler on the PDP-10 (replacing the
third-party F4/F40) was written in BLISS-10, and continued to be supported
through 11 versions (uh, 9, for the octal-impaired), which required that
every PDP-10 running Fortran have a copy of BLISS-10 in order to handle SPR
patches.

Then there was BLISS-11, which was a cross-compiler for the PDP-11 which ran
on the PDP-10.

The later Bliss-36 was done by the VMS language folks, along with Bliss-32
and Bliss-16.  When Digital made the latter two compilers available for free
to the relevant customers, I got them to release Bliss-36 to us at XKL for
the 36-bit crowd.

-- 
Rich Alderson                                   news@alderson.users.panix.com
    the russet leaves of an autumn oak/inspire once again the failed poet/
    to take up his pen/and essay to place his meagre words upon the page...

m.kraemer@gsi.de (Michael Kraemer) writes:

> In article <cxJeWhYXnsCo@eisner.encompasserve.org>,
> koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:

>> In article <i8fchd$sf$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de
>> (Michael Kraemer) writes:

>>> since nobody uses it outside of the VMS niche,
>>> this is a bit hard to verify.

>>    Since BLISS was invented on a PDP-10,

> gone.

That would be news to my former employer, which uses a PDP-10 processor on
an FPGA in every one of their products.

>>  ported to VAX, Alpha, and
>>    Itanium, 

> VMS niche, as said.

>> targeted PDP-11 

> gone.

News to the folks at XX2247, successors of Mentec.

>> (via cross compilers), and is being
>>    ported to IA-32,

> in other words: not available, let alone being used.

>> the claim that it is highly portable it's fairly easy 
>>    to verify.

> ported to mostly irrelevant platforms,
> which leaves VMS, as I said.

In your highly uninformed opinion.

>>    If you don't know how different PDP-10 was from the rest of those,
>>    shut up and go learn.

> should I care, except for historical reasons? 

Yes.

-- 
Rich Alderson                                   news@alderson.users.panix.com
    the russet leaves of an autumn oak/inspire once again the failed poet/
    to take up his pen/and essay to place his meagre words upon the page...

glen herrmannsfeldt <gah@ugcs.caltech.edu> writes:

> Bill Gunshannon <billg999@cs.uofs.edu> wrote:
>> In article <i8dak6$5h9$1@usenet01.boi.hp.com>,
>>        "FredK" <fred.nospam@dec.com> writes:
> (snip)

>>> Bill, I didn't mean to imply UNIX or C "invented" zwro terminated strings. 

>> Not singling you out.  It is just a common thread here where people
>> seem to ignore the fact that Unix/C merely picked up on a method that
>> was pretty much standard accross the industry.

> As far as I know, it was not commonly used by IBM at the time.
> Pretty much the IBM code from OS/360 and successors uses count
> fields with strings, even text strings.  (EBCDIC also has a NUL
> at position X'00', so it could have been done.)

The 360 architecture *requires* counts in the instructions which handle
strings.

>>> C made it the language standard, 

>> De facto made it a standard.  To the best of my knowledge C 
>> does not have a string data type at all. 

>>>           UNIX enshrined it in it's interfaces. 

> Isn't C older than unix?  In any case, Unix and C are closely
> related, and seem to have helped propagate it.

No, C came along in order to facilitate the port of Unix from the PDP-7
(18-bit word address architecture) to the PDP-11.

> Otherwise, it seems to me that DEC used it before C.

Yes, on the PDP-6 and PDP-10, and I believe on the 12- and 18-bit machines.
Probably on the PDP-11 as well, before Unix came along.

-- 
Rich Alderson                                   news@alderson.users.panix.com
    the russet leaves of an autumn oak/inspire once again the failed poet/
    to take up his pen/and essay to place his meagre words upon the page...

On 04-10-2010 02:15, smithfarm wrote:
> This whole troll-fest --er, I meant to say "discussion" -- got me
> thinking about one thing: the mindset that says "VMS is not good for
> <this or that>". On the one hand, it is true that VMS's engineering
> resources are limited by factors such as the dwindling number of
> qualified engineers and the number of hours they spend prattering --er,
> I meant to say "recovering" -- on c.o.v. so the question of where to
> allocate those resources is a valid one.
>
> On the other hand, if VMS is ever to gain any traction as a viable "real
> world" alternative to Windows and Linux, there is a certain minimal (and
> shifting) feature set that it must support. What kinds of things that
> would include is debatable, but certainly a web browser (and, yes,
> something that doesn't choke on quoted-pukeable too) would be high up on
> my list.
>
> Having to run and maintain Linux and Windows boxen alongside VMS systems
> just because VMS lacks "this or that" is sub-optimal. VMS is a
> generalized operating system. Anything that can be done on Windows or
> Linux *can* be done on VMS. Question is, will it be done. And that's a
> resource-allocation question.

Anything is possible with sufficient resources.

10 B$ could change everything.

But it will not happen.

And if it did then I don't think we would like
the result!

Arne

On Tue, 05 Oct 2010 23:51:00 -0700, H Vlems wrote:

>> I do remember in the OS/360 days putting a carriage return in a literal
>> constant in Fortran.
>>
>> -- glen
> 
> No, unix is definitely older than C. When unix was developed on the
> first platform, a bare PDP-7,
> they used assembly language to build the os. The next version was
> rewritten in a language that
> derived from BCPL (a British invention). The port to the PDP-11 saw an
> upgrade to this and that
> became known as C.

Actually, there was a language 'in between' called B.

Funnily enough, BCPL used counted strings....



-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

In article <4cacee83$0$15417$c3e8da3$cc4fe22d@news.astraweb.com>,
	JF Mezei <jfmezei.spamnot@vaxination.ca> writes:
> Bill Gunshannon wrote:
> 
>> Oh please.  Apple still makes up a very small part of the true Unix
>> market 
> 
> and later on:
> 
> 
>> "It's the applications  stupid!" (Not attacking, just paraphrasing something
>> most people will probably remember. :-)
> 
> 
> You should know that OS-X is more Unix than you think.

What are you yammering about.  I know that OS-X is 100% Unix.

> 
> With support for X11, GTK etc, OS-X is in fact just as Unixy a the guy
> next door. The Apple servers come with standard Unix applications
> (bind9, postfix, dovecot, jabber, ftp, etc etc). And you can "make" just
> about any application designed for unix on a Mac.
> 
> 
> So, in the "its the applications stupid!" department, OS-X is very much
> Unix and, when you combine that with the OS-X "native" applications that
> use the Apple proprietary frameworks and COCOA UI, you get a fairly
> impressive application portfolio.
> 
> Windows does not come out of the box with unix compatibility.

Except that we were talking about why VMS changing hands was not going
to make it like a Mac.  Mac already has all the applications that are
available for Unix and VMS has ---- None.  You really need to try to
follow a discussion more.  Especially when you are them going to cut
all the relevant text out of your comments.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

On 04-10-2010 08:31, Michael Kraemer wrote:
> In article<4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com>, JF Mezei
> <jfmezei.spamnot@vaxination.ca>  writes:
>>
>> Also, if you need another OS to manage a server OS, you might as well
>> use that other OS as your server OS.
>
> exactly.
> Moreover, to come up with another of those popular
> car analogies: what some people suggest here is
> to sell a car restricted to engine and tank ("server-centric").
> If you want to actually drive it,
> you'd have to buy/use a second one,
> which will give you seats, tires and a driving wheel,
> so you can hook up the first one.
> I think most, if not all customers would just
> buy the second, complete one, even if it
> has a "weaker" (by whatever measure) engine
> than the first one.

That analogy does not seem very relevant.

It seems a bit more relevant to compare with people
wanting a truck to transport tons of sand and a
sedan to visit the family in.

Arne

On 04-10-2010 04:18, JF Mezei wrote:
> Also, if you need another OS to manage a server OS, you might as well
> use that other OS as your server OS.

That does not seem to be that common in the industry.

z/OS, i, AIX, SOlaris, HP-UX seems to be sold only for server
usage.

Windows is sold for multiple purposes, but the difference
between the versions are getting bigger over time. NT 4.0 Workstation
and NT 4.0 server were really the same OS - Windows 7, Windows 2008R2
and Windows Phone 7 are quite different.

Same with Linux. RHEL, Ubuntu and Android  use the same kernel,
but are quite different.

MacOS X may be the one today with the smallest difference between
server and desktop, but I don't know.

Arne

Rich Alderson <news@alderson.users.panix.com> wrote:
(snip, I wrote)
 
>> As far as I know, it was not commonly used by IBM at the time.
>> Pretty much the IBM code from OS/360 and successors uses count
>> fields with strings, even text strings.  (EBCDIC also has a NUL
>> at position X'00', so it could have been done.)
 
> The 360 architecture *requires* counts in the instructions 
> which handle strings.

Well, you could loop over them with an index register.

There is also TRT which can search though a string and stop
at a specified byte.  (It also has a length, but can stop
early.)

Otherwise, they have now added instructions to do C-like
string operations in z/Architecture.

-- glen

Rich Alderson <news@alderson.users.panix.com> wrote:
(snip)
 
> In point of fact, the later Fortran compiler on the PDP-10 (replacing the
> third-party F4/F40) was written in BLISS-10, and continued to be supported
> through 11 versions (uh, 9, for the octal-impaired), which required that
> every PDP-10 running Fortran have a copy of BLISS-10 in order to handle SPR
> patches.

Version numbers are in octal!  I know that IBM serial numbers
are in hex, but I believe version numbers for software are
in decimal.

-- glen

On 2010-10-06, Rich Alderson <news@alderson.users.panix.com> wrote:
>
> News to the folks at XX2247, successors of Mentec.
>

What exactly do you mean by Mentec successor ?

Is this a successor in the legal sense of the word or a unrelated
organisation ? Are they resuming Mentec's PDP-11 OS operations ?

A search on Google didn't reveal anything apart from a website with
a picture of a key.

Thanks for any information you may have,

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980's technology to a 21st century world

In article <4cad2973$0$23758$14726298@news.sunsite.dk>,
	Arne Vajh�j <arne@vajhoej.dk> writes:
> On 04-10-2010 04:18, JF Mezei wrote:
>> Also, if you need another OS to manage a server OS, you might as well
>> use that other OS as your server OS.
> 
> That does not seem to be that common in the industry.
> 
> z/OS, i, AIX, SOlaris, HP-UX seems to be sold only for server
> usage.

Huh??   I have been sold and used both HP-UX and Solaris as desktop OSes.

> 
> Windows is sold for multiple purposes, but the difference
> between the versions are getting bigger over time. NT 4.0 Workstation
> and NT 4.0 server were really the same OS - Windows 7, Windows 2008R2
> and Windows Phone 7 are quite different.
> 
> Same with Linux. RHEL, Ubuntu and Android  use the same kernel,
> but are quite different.

Only difference I have ever seenin Linux distros is the install.  They
are all the same under the hood.

> 
> MacOS X may be the one today with the smallest difference between
> server and desktop, but I don't know.

Like the others I mentioned above, there is no difference between server
and desktop once you look under the hood.

bill


-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <8h5u0aFiciU1@mid.individual.net>, billg999@cs.uofs.edu (Bill
Gunshannon) writes:
> In article <4cad2973$0$23758$14726298@news.sunsite.dk>,
> 	Arne Vajh�j <arne@vajhoej.dk> writes:
> > On 04-10-2010 04:18, JF Mezei wrote:
> >> Also, if you need another OS to manage a server OS, you might as well
> >> use that other OS as your server OS.
> > 
> > That does not seem to be that common in the industry.
> > 
> > z/OS, i, AIX, SOlaris, HP-UX seems to be sold only for server
> > usage.
> 
> Huh??   I have been sold and used both HP-UX and Solaris as desktop OSes.

you may add AIX here.
But all three vendors have dropped their workstation lines
in the past three or four years, with IBM being the last one
(end 2008 IIRC).
 
> Like the others I mentioned above, there is no difference between server
> and desktop once you look under the hood.

except that in a "server only" environment the vendor will
drop support of X servers and/or gfx cards and associated sw
sooner or later. Has already happened for HP-UX at least.

In article <4cad2973$0$23758$14726298@news.sunsite.dk>,
=?ISO-8859-1?Q?Arne_Vajh=F8j?= <arne@vajhoej.dk> writes:
> On 04-10-2010 04:18, JF Mezei wrote:
> > Also, if you need another OS to manage a server OS, you might as well
> > use that other OS as your server OS.
> 
> That does not seem to be that common in the industry.
> 
> z/OS, i, AIX, SOlaris, HP-UX seems to be sold only for server
> usage.

For the Unices there are still a couple of desktop systems
in use and supported by the OS's.
However, the deemphasis of desktops/entry level 
may well give them the same kind of problems VMS has had for
the past two decades, i.e.
no desktop presence -> no awareness -> less developers
-> less software -> less customers.
But at least the people there don't complain about
web browsers being available.

In article <i8kiih$c7d$1@lnx107.hrz.tu-darmstadt.de>,
	m.kraemer@gsi.de (Michael Kraemer) writes:
> In article <8h5u0aFiciU1@mid.individual.net>, billg999@cs.uofs.edu (Bill
> Gunshannon) writes:
>> In article <4cad2973$0$23758$14726298@news.sunsite.dk>,
>> 	Arne Vajh�j <arne@vajhoej.dk> writes:
>> > On 04-10-2010 04:18, JF Mezei wrote:
>> >> Also, if you need another OS to manage a server OS, you might as well
>> >> use that other OS as your server OS.
>> > 
>> > That does not seem to be that common in the industry.
>> > 
>> > z/OS, i, AIX, SOlaris, HP-UX seems to be sold only for server
>> > usage.
>> 
>> Huh??   I have been sold and used both HP-UX and Solaris as desktop OSes.
> 
> you may add AIX here.
> But all three vendors have dropped their workstation lines
> in the past three or four years, with IBM being the last one
> (end 2008 IIRC).
>  
>> Like the others I mentioned above, there is no difference between server
>> and desktop once you look under the hood.
> 
> except that in a "server only" environment the vendor will
> drop support of X servers and/or gfx cards and associated sw
> sooner or later. Has already happened for HP-UX at least.

Interesting idea as a quick search of HP-UX by Google returned
the following:
--------------------------------------------------

HP-UX desktop announcement
HP AllianceONE Partner Program 

HP will continue to deliver CDE as the default desktop for HP-UX.

HP has released an updated HP-UX Desktop Developer's Toolkit for 11i v2
with additional support for the HP Itanium desktop development environment.
The symbolic links to HP Itanium shared libraries and header files are
delivered as part of this toolkit. HP Itanium shared libraries for CDE,
Audio, and Imaging are delivered via patches.

 More information

An unsupported port of GNOME v1.4 is available for HP-UX on both the
PA-RISC and HP Itanium platforms. 

 More information

HP currently has no plans for GNOME 2.x on HP-UX.
--------------------------------------------------

And then we have this Solaris announcement from Oracle, their new owner:

--------------------------------------------------

http://www.sun.com/software/solaris/javadesktopsystem.xml

Sun Java Desktop System, Release 3 for the Solaris 10 Operating System      
  
 

Introducing Java Desktop System, 

Release 3 for the Solaris 10 Operating System
With Java Desktop System, Release 3 for Solaris 10, Sun offers customers
the combination of a modern, intuitive desktop environment and the
advanced feature set and rock-solid security of the Solaris OS. Solaris
users benefit from improved functionality and greater user-friendliness
through Java Desktop System. Java Desktop System delivers the best in
open source software with the technical innovation and release discipline
in a fully integrated, comprehensive desktop. For a list of features for
Java Desktop System, Release 3 for Solaris 10 OS, go here.
 
--------------------------------------------------

And then we come to AIX:

--------------------------------------------------

http://www-03.ibm.com/systems/power/software/aix/sysmgmt/desktop/index.html


Desktop

CDE   The Common Desktop Environment (CDE) is a graphical user interface
that serves as the default desktop in AIX systems. CDE, which you can
customize, provides services to system administrators, application
developers, and end users in a consistent fashion across multiple
platforms. CDE allows you to open multiple desktops in a highly reliable
and stable environment. Each desktop can then run multiple independent
applications without interfering with the operations of the other three.

KDE    The K Desktop Environment (KDE) is also available as part of Linux
Affinity and the free AIX Toolbox for Linux Applications. This open-source
graphical desktop is built on top of the X11 environment for UNIX
operating systems.

KDE contains a compound application development framework, which hosts a
large collection of applications. The KDE-Graphics application features
Kpaint, Kdvi, KGhostview, and Kfax. KDE-Admin provides the tools needed
for system administrators, while Klibs provides various run-time libraries
such as kdcore, kdeui, and khtwm. Opportunities for customization abound
with the extremely flexible KDE user interface.

GNOME    The GNU Network Object Model Environment (GNOME) desktop is also
available as part of Linux Affinity and the free AIX Toolbox for Linux
Applications (ATLA). GNOME.a powerful application framework for software
development.is highly configurable, allowing you to set the desktop's look
and feel to suit your preference.

Applications available for use with GNOME include Qt, a cross-platform
C++ application framework; GLib, a library for the C language containing
portability and utility functions; and GTK, a container-based tool kit.

--------------------------------------------------

Don't now where you got your info, but it seems to conflict with the
current plans of all three venders.



bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <mddk4lvhsqz.fsf@panix5.panix.com>, Rich Alderson <news@alderson.users.panix.com> writes:
> 
> In point of fact, the later Fortran compiler on the PDP-10 (replacing the
> third-party F4/F40) was written in BLISS-10, and continued to be supported
> through 11 versions (uh, 9, for the octal-impaired), which required that
> every PDP-10 running Fortran have a copy of BLISS-10 in order to handle SPR
> patches.

   Although I had BLISS-10 loaded on my 2050 and 2060 so I didn't have
   to write too much Macro-10, I don't recall any Fortran patches that
   weren't in binary.  But BLISS-10 was free and BLISS-36 wasn't, which
   would explain why when they rewrote the Fortran compiler in BLISS,
   they didn't use BLISS-36, which they already had.

> 
> The later Bliss-36 was done by the VMS language folks, along with Bliss-32
> and Bliss-16.  When Digital made the latter two compilers available for free
> to the relevant customers, I got them to release Bliss-36 to us at XKL for
> the 36-bit crowd.

   Gee, I wish I'd have had BLISS-36 and -16.  But I didn't get BLISS-32 
   until DEC made it free, years after I left the shop with PDP-10.

In article <8h6363Ffh2U1@mid.individual.net>, billg999@cs.uofs.edu (Bill
Gunshannon) writes:
> 
> Don't now where you got your info, but it seems to conflict with the
> current plans of all three venders.

Not so fast. The difference is in the hardware, not software.
The X client libs, CDE & friends will still be there, so you'd
still be able to run and develop typical "desktop" apps,
as long as you have the Unix box and some X server device.
Traditionally the latter two were combined in a classical workstation.
This soon will no longer be supported, for sure in the higher releases
of HP-UX, don't know yet about AIX >= 7.
Of course one could use any X capable thin client connected to a
server box, but there the low end is de-emphasized as well,
which begs the question whether one has to buy a Superdome class machine just
to port a web browser or similar.

 

In article <mddeic3hrj2.fsf@panix5.panix.com>, Rich Alderson <news@alderson.users.panix.com> writes:
> 
> The 360 architecture *requires* counts in the instructions which handle
> strings.

   But I could write null-terminated string handling using byte
   instructions, couldn't I.  (I don't have my green card handy to look
   them up.)

   VAX has MOVx instructions which require a count, too, but
   LOCC can figure it out, you just have to burn up the extra CPU
   time to do the LOCC.  I was suprized how many VAX C RTL string
   routines started with LOCC when I thought they could have
   got away with MOVTUC and a page of null translation table.

"Michael Kraemer" <m.kraemer@gsi.de> wrote in message 
news:i8kmrj$ebh$1@lnx107.hrz.tu-darmstadt.de...
> In article <8h6363Ffh2U1@mid.individual.net>, billg999@cs.uofs.edu (Bill
> Gunshannon) writes:
>>
>> Don't now where you got your info, but it seems to conflict with the
>> current plans of all three venders.
>
> Not so fast. The difference is in the hardware, not software.
> The X client libs, CDE & friends will still be there, so you'd
> still be able to run and develop typical "desktop" apps,
> as long as you have the Unix box and some X server device.
> Traditionally the latter two were combined in a classical workstation.
> This soon will no longer be supported, for sure in the higher releases
> of HP-UX, don't know yet about AIX >= 7.
> Of course one could use any X capable thin client connected to a
> server box, but there the low end is de-emphasized as well,
> which begs the question whether one has to buy a Superdome class machine 
> just
> to port a web browser or similar.
>
>

This is both true and false.  Most Integrity systems (the exception being 
the cellular systems) have an integrated 2D graphics chip (currently an AMD 
RN50) which HP-UX and VMS support for X11 graphics alomg with USB ports for 
KB and mice.  There are also add-in cards for multi-head.

What is true is that there is no "workstation" (or system designed for the 
desktop) - or high performance 3D cards.

What is also true is that with the lack of a "desktop" system, some of the 
choices are to use the smallest server available as a deskside system (like 
the RX2620) OR to use a thin client that runs X11.

One can port a web browser on anything from a RX2620 to a Superdome.  One 
can develop it with the built in graphics or remote to  thin client.

Note that a "thin client" could be a Linux workstation as well.

In article <i8kmrj$ebh$1@lnx107.hrz.tu-darmstadt.de>,
	m.kraemer@gsi.de (Michael Kraemer) writes:
> In article <8h6363Ffh2U1@mid.individual.net>, billg999@cs.uofs.edu (Bill
> Gunshannon) writes:
>> 
>> Don't now where you got your info, but it seems to conflict with the
>> current plans of all three venders.
> 
> Not so fast. The difference is in the hardware, not software.

Yeah.  Solaris runs on PC's as well as Sparc (and I still have a perfectly
functioning Sparc Desktop System at home).  HP-UX, as far as i know, only
runs on PA-RISC and Itanium.  And, AIX (again, as far as I know) only on
Power.  So, Solaris's PC product will continue to have sound and video
capability.  But what of the others.  I can't speak for HP as their reput-
ation in this regard leaves much to be desired but the other two have
never been known to announce new products or continued support immediately
before terminating a product.

> The X client libs, CDE & friends will still be there, so you'd
> still be able to run and develop typical "desktop" apps,
> as long as you have the Unix box and some X server device.
> Traditionally the latter two were combined in a classical workstation.
> This soon will no longer be supported, for sure in the higher releases
> of HP-UX, don't know yet about AIX >= 7.

Like I said above, I have no failth in HP but if the others are announcing
products today I do not expect them to reverse themselves tomorrow.

> Of course one could use any X capable thin client connected to a
> server box, but there the low end is de-emphasized as well,
> which begs the question whether one has to buy a Superdome class machine just
> to port a web browser or similar.
 
That is, of course, true.  While most of us old timers thought we saw the
X-terminal fade into oblivion, it actually seems to be back as most of
the thin-clients I look at today are also  true X-terminals.  

Oof course, all the announcements I read mentioned audio as well as
video and we all know the audio support in X-11 never went far enough
to be considered a practical product.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

On Thu, 2010-10-07 at 15:40 +0000, Bill Gunshannon wrote:
> Oof course, all the announcements I read mentioned audio as well as
> video and we all know the audio support in X-11 never went far enough
> to be considered a practical product.=20

Using Linux as a thin client will also provide you with audio and video
capabilities on OpenVMS.=20

Personally, I think the DECwindows Motif CDE looks very outdated next to
a modern Gnome 3.0 desktop. Perhaps it would be an interesting exercise
to port Gnome 3.0 to the OpenVMS architecture.=20
--=20
Tactical Nuclear Kittens

In article <vn81o7-cfk.ln1@nntp.local.net>, Captain Pugwash
<alex.buell@munted.org.uk> writes:
> On Thu, 2010-10-07 at 15:40 +0000, Bill Gunshannon wrote:
> > Oof course, all the announcements I read mentioned audio as well as
> > video and we all know the audio support in X-11 never went far enough
> > to be considered a practical product.=20
> 
> Using Linux as a thin client will also provide you with audio and video
> capabilities on OpenVMS.=20

and what's again the name of the vendor-supported protocol
which transports audio data from VMS to your Linux box?
Audio11?
 

In article <i8kitk$ca1$1@lnx107.hrz.tu-darmstadt.de>, m.kraemer@gsi.de (Michael Kraemer) writes:
> For the Unices there are still a couple of desktop systems
> in use and supported by the OS's.
> However, the deemphasis of desktops/entry level 
> may well give them the same kind of problems VMS has had for
> the past two decades, i.e.
> no desktop presence

   What, no OpenOffice on a Sun?  Saints preserve us!

In article <vn81o7-cfk.ln1@nntp.local.net>, Captain Pugwash <alex.buell@munted.org.uk> writes:
> 
> Personally, I think the DECwindows Motif CDE looks very outdated next to
> a modern Gnome 3.0 desktop. Perhaps it would be an interesting exercise
> to port Gnome 3.0 to the OpenVMS architecture.=20

   Personnaly, I think both Gnome and KDE owe too much influence
   to Windows.

   But grab the source of either one of those and be our guest.

On Thu, 2010-10-07 at 13:14 -0600, Bob Koehler wrote:
> > Personally, I think the DECwindows Motif CDE looks very outdated
> next to
> > a modern Gnome 3.0 desktop. Perhaps it would be an interesting
> exercise
> > to port Gnome 3.0 to the OpenVMS architecture.=3D20
>=20
>    Personnaly, I think both Gnome and KDE owe too much influence
>    to Windows.
>=20
>    But grab the source of either one of those and be our guest.=20

Yeah, one of these days I shall. I already have the sources.=20
--=20
Tactical Nuclear Kittens

"Bill Gunshannon" <billg999@cs.uofs.edu> wrote in message 
news:8h3vo2Fv5jU2@mid.individual.net...
> In article <poKUlBa6k0b9@eisner.encompasserve.org>,
> koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:
>>  On 10/06/2010 09:35 PM, Bill Gunshannon wrote:
>>> Given that people have found it not prohibitively difficult to either
>>> graft front-ends onto the Gun Compiler or write translators to C (often
>>> as a precursor to a later GCC Ffrontend) has anyone ever given any
>>> consideration to makeing a freely available Bliss or DIBOL compiler?
>>
>>    The BLISS compiler for VMS has been freely available since soon
>>    after VMS was ported to Alpha.
>>
>>    There is a free BLISS compiler being developed in relationship
>>    too FreeVMS.  You can get early releases, including the source.
>
> If you read what I wrote, you would have seen that I am interested in
> writting Bliss porgrams for VMS.  I already know that works.  I am
> interested in seeing if it is possible to create a version of Unix
> that lacks even the few small security problems that currently may
> exist as a result of it being written in C. (Not saying there are any
> and there really are none I know of that haven't been addressed in
> the C version.  Just have never garnered any interest among the unix
> community.)
>>
>>    There is a source for a PDP-11 BLISS compiler freely available,
>>    IIRC it's a cross compiler meant to run on VMS.
>
> Now that I would be interested in.  The only PDP-11 Bliss compiler
> I was aware of was for the PDP-10 and was not available in source
> anywhere I could find it.  Given a Bliss compiler in source to start
> with I might try writting a Bliss to C translator myself (in my
> copious freetime. :-)  Unless, of course it is like the current state
> of "free" ADA.
>
> bill
>
>
>
> -- 
> Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
> billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
> University of Scranton   |
> Scranton, Pennsylvania   |         #include <std.disclaimer.h>

"the current state of "free" ADA."

It's Ada. Not ADA. Named after Ada Lovelace.

And what exactly does Bill see as the state of "free" Ada?

What most of the world sees is that companies like Boeing, Airbus, and their 
suppliers, rely on free (as in open source, as in freely downloadable) Ada 
(GNAT, the gcc variant which supports Ada) or its commercially supported big 
brother from AdaCore, GNAT Pro. Same gcc source code whichever one you pick 
(give or take a few months bugfixing, as their GPL one is released once or 
twice a year, I forget, whereas the commercial one presumably gets bugfixes 
earlier).

GNAT even has VMS as a commercially supported platform (probably not 
unrelated to DEC's decision not to develop an Ada 9x compiler).

For those still supporting projects with MC68K as target (these embedded 
things hang around a lot longer than the average IT project, many have been 
around longer than many IT people have been working), there is still 
XD-Ada - the DEC Ada front end, with an MC68K back end, running on VMS, 
originally co-developed by DEC's Ada compiler team and SD-Scicon's 68K 
folks. Still available and supported on VAX VMS, on Alpha VMS, and on 
Integrity VMS. Not free, but well regarded.

The usual search engines will find everything the well-informed reader might 
want to know about GNAT or XD Ada. 

In article <i8kt1f$hda$1@lnx107.hrz.tu-darmstadt.de>,
	m.kraemer@gsi.de (Michael Kraemer) writes:
> In article <vn81o7-cfk.ln1@nntp.local.net>, Captain Pugwash
> <alex.buell@munted.org.uk> writes:
>> On Thu, 2010-10-07 at 15:40 +0000, Bill Gunshannon wrote:
>> > Oof course, all the announcements I read mentioned audio as well as
>> > video and we all know the audio support in X-11 never went far enough
>> > to be considered a practical product.=20
>> 
>> Using Linux as a thin client will also provide you with audio and video
>> capabilities on OpenVMS.=20
> 
> and what's again the name of the vendor-supported protocol
> which transports audio data from VMS to your Linux box?
> Audio11?

Sadly, there actually was audio intended in X11 but I know of no one
who did it in a standard way.  I have a couple of HP (Pre-VMS days)
X-terminals that had an available audio option but I have never been
able to find any of the add-on boards so I have no idea how or even
if they actually could do audio from X.

bill
  

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

Bill Gunshannon wrote:

> Oof course, all the announcements I read mentioned audio as well as
> video and we all know the audio support in X-11 never went far enough
> to be considered a practical product.

Since Linux uses X11, and since Linux supports applicatiosn such as VLC
which play movies with surround sound, X11's limited sound capabilities
don't seem to stop Linux.

In article <8h6lecF83rU1@mid.individual.net>, billg999@cs.uofs.edu (Bill
Gunshannon) writes:
> 
> Sadly, there actually was audio intended in X11 but I know of no one
> who did it in a standard way.  I have a couple of HP (Pre-VMS days)
> X-terminals that had an available audio option but I have never been
> able to find any of the add-on boards so I have no idea how or even
> if they actually could do audio from X.
> 

Not really "audio", but there's this fun demo, "xtune",
which at least sounds the bell so one can play simple tunes
via a note string.  

Bob Koehler wrote:

>    Personnaly, I think both Gnome and KDE owe too much influence
>    to Windows.
> 
>    But grab the source of either one of those and be our guest.

Does VMS now have a current version of X11 ? Last I checked, it ran an
older version of it which lacks some of the functionality needed by
modern apps.

In article <4cae2435$0$19547$c3e8da3$2e0018d8@news.astraweb.com>, JF Mezei
<jfmezei.spamnot@vaxination.ca> writes:
> Bill Gunshannon wrote:
> 
> > Oof course, all the announcements I read mentioned audio as well as
> > video and we all know the audio support in X-11 never went far enough
> > to be considered a practical product.
> 
> Since Linux uses X11, and since Linux supports applicatiosn such as VLC
> which play movies with surround sound, X11's limited sound capabilities
> don't seem to stop Linux.

It won't stop classical workstations with a decent audio card either,
since the audio data go directly to the card, not via the X protocol.
X defines pixmaps etc but not "audiomaps", AFAIK, so "remote audio"
is not possible.

On Thu, 2010-10-07 at 19:58 +0000, Michael Kraemer wrote:
> > Since Linux uses X11, and since Linux supports applicatiosn such as
> VLC
> > which play movies with surround sound, X11's limited sound
> capabilities
> > don't seem to stop Linux.
>=20
> It won't stop classical workstations with a decent audio card either,
> since the audio data go directly to the card, not via the X protocol.
> X defines pixmaps etc but not "audiomaps", AFAIK, so "remote audio"
> is not possible.=20

With Gnome or KDE, you can direct audio to be played remotely on the
thin client if it can support it.=20
--=20
Tactical Nuclear Kittens

On Thu, 2010-10-07 at 15:54 -0400, JF Mezei wrote:
> Bob Koehler wrote:
>=20
> >    Personnaly, I think both Gnome and KDE owe too much influence
> >    to Windows.
> >=20
> >    But grab the source of either one of those and be our guest.
>=20
> Does VMS now have a current version of X11 ? Last I checked, it ran an
> older version of it which lacks some of the functionality needed by
> modern apps.

Probably not. Maybe through a 3rd party add-on?=20
--=20
Tactical Nuclear Kittens

"Arne Vajh�j" <arne@vajhoej.dk> wrote in message 
news:4cad287d$0$23751$14726298@news.sunsite.dk...
> On 04-10-2010 08:31, Michael Kraemer wrote:
>> In article<4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com>, JF Mezei
>> <jfmezei.spamnot@vaxination.ca>  writes:
>>>
>>> Also, if you need another OS to manage a server OS, you might as well
>>> use that other OS as your server OS.
>>
>> exactly.
>> Moreover, to come up with another of those popular
>> car analogies: what some people suggest here is
>> to sell a car restricted to engine and tank ("server-centric").
>> If you want to actually drive it,
>> you'd have to buy/use a second one,
>> which will give you seats, tires and a driving wheel,
>> so you can hook up the first one.
>> I think most, if not all customers would just
>> buy the second, complete one, even if it
>> has a "weaker" (by whatever measure) engine
>> than the first one.
>
> That analogy does not seem very relevant.
>
> It seems a bit more relevant to compare with people
> wanting a truck to transport tons of sand and a
> sedan to visit the family in.
>
> Arne
>

I like this analogy. The different versions of Windows are largely like 
buying different versions of the same basic car with different levels of 
trim, maybe with a 4x4 version for the really adventurous. As you point out, 
in the real IT world, you don't use a Ford Escort when a 7 tonne truck is 
needed, or vice versa. Windows Mobile is perhaps a pedal cycle in this 
analogy.

Either way, afaict, no other field of technology or engineering or craft 
would be willing to accept the same level of vendor-dependent monoculture as 
the IT world routinely accepts. 

Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP> writes:

> On 2010-10-06, Rich Alderson <news@alderson.users.panix.com> wrote:

>> News to the folks at XX2247, successors of Mentec.

> What exactly do you mean by Mentec successor ?

> Is this a successor in the legal sense of the word or a unrelated
> organisation ? Are they resuming Mentec's PDP-11 OS operations ?

Yes, a legal successor.

I thought they would have said something for themselves by now, at least on
their site.  I shouldn't have said anything, I guess.

-- 
Rich Alderson                                   news@alderson.users.panix.com
    the russet leaves of an autumn oak/inspire once again the failed poet/
    to take up his pen/and essay to place his meagre words upon the page...

glen herrmannsfeldt <gah@ugcs.caltech.edu> writes:

> Rich Alderson <news@alderson.users.panix.com> wrote:
> (snip)

>> In point of fact, the later Fortran compiler on the PDP-10 (replacing the
>> third-party F4/F40) was written in BLISS-10, and continued to be supported
>> through 11 versions (uh, 9, for the octal-impaired), which required that
>> every PDP-10 running Fortran have a copy of BLISS-10 in order to handle SPR
>> patches.

> Version numbers are in octal!  I know that IBM serial numbers
> are in hex, but I believe version numbers for software are
> in decimal.

I give you my word that I installed the following new versions of Fortran-20
as they came out in succession:  v5, v6, v7, v10, and v11.  There was a lot
of discussion among the 36-bit crowd when v10 followed v7.

Most version numbers are in decimal for DEC software, too.  Then again, the
base-40 notation used in symbol tables is called RADIX50 in DEC assemblers
and debuggers.

-- 
Rich Alderson                                   news@alderson.users.panix.com
    the russet leaves of an autumn oak/inspire once again the failed poet/
    to take up his pen/and essay to place his meagre words upon the page...

In article <4cae2435$0$19547$c3e8da3$2e0018d8@news.astraweb.com>, JF Mezei <jfmezei.spamnot@vaxination.ca> writes:
> 
> Since Linux uses X11, and since Linux supports applicatiosn such as VLC
> which play movies with surround sound, X11's limited sound capabilities
> don't seem to stop Linux.

   Just because an OS uses X11 for graphics and keyboard does not
   mean it has an X11 based audio.

   I don't ever recall audio in the X11 API.

In article <4cae2584$0$19547$c3e8da3$2e0018d8@news.astraweb.com>, JF Mezei <jfmezei.spamnot@vaxination.ca> writes:
> 
> Does VMS now have a current version of X11 ? Last I checked, it ran an
> older version of it which lacks some of the functionality needed by
> modern apps.

   Alpha and Itanium run pretty much unfettered copies of the available
   X11, so you can port a later one of those, too.

   But don't try it on VAX and expect to have any performance.

   OBTW, you will have to modify xAppGetInput, or whatever its called
   now.

"Bob Koehler" <koehler@eisner.nospam.encompasserve.org> wrote in message 
news:bu3yVeRZLbuQ@eisner.encompasserve.org...
> In article <4cae2584$0$19547$c3e8da3$2e0018d8@news.astraweb.com>, JF Mezei 
> <jfmezei.spamnot@vaxination.ca> writes:
>>
>> Does VMS now have a current version of X11 ? Last I checked, it ran an
>> older version of it which lacks some of the functionality needed by
>> modern apps.
>
>   Alpha and Itanium run pretty much unfettered copies of the available
>   X11, so you can port a later one of those, too.
>
>   But don't try it on VAX and expect to have any performance.
>
>   OBTW, you will have to modify xAppGetInput, or whatever its called
>   now.

VMS uses the X11 stream that came out of MIT - I think X Open owns it now 
(can't remember).  Linux uses xFree86.  Both streams are in use, the older 
one changes less, the xFree86 one changes often.

There are many modifications in the libraries for VMS, but for the core 
X11 - most of the OS-specific stuff is in the place where OSes have their 
own stuff (in our case OS_VMS) - interfaces to drivers etc.  Motif also has 
a OS-specific component to to things like access files...  but the library 
code isn't quite so clean and little changes happen everywhere (lots of 
assumptions about file strings).  But if you don't mind UNIX path strings 
and such, most of Motif/Xlib will port without too much trouble.

"Bob Koehler" <koehler@eisner.nospam.encompasserve.org> wrote in message 
news:b+1fU6vUu4EA@eisner.encompasserve.org...
> In article <4cae2435$0$19547$c3e8da3$2e0018d8@news.astraweb.com>, JF Mezei 
> <jfmezei.spamnot@vaxination.ca> writes:
>>
>> Since Linux uses X11, and since Linux supports applicatiosn such as VLC
>> which play movies with surround sound, X11's limited sound capabilities
>> don't seem to stop Linux.
>
>   Just because an OS uses X11 for graphics and keyboard does not
>   mean it has an X11 based audio.
>
>   I don't ever recall audio in the X11 API.
>

It would be done as an extension.  X11 would probably be a bad transport for 
it.  Plus if what you really want is video, you need to keep the audio and 
video streams in sync.

In article <4cae2435$0$19547$c3e8da3$2e0018d8@news.astraweb.com>,
	JF Mezei <jfmezei.spamnot@vaxination.ca> writes:
> Bill Gunshannon wrote:
> 
>> Oof course, all the announcements I read mentioned audio as well as
>> video and we all know the audio support in X-11 never went far enough
>> to be considered a practical product.
> 
> Since Linux uses X11, and since Linux supports applicatiosn such as VLC
> which play movies with surround sound, X11's limited sound capabilities
> don't seem to stop Linux.

Once again you are missing the point of the discussion.  It was about
X-terminals ala Thin-Clients doing audio remotely, not running X on
the host system (which we know VMS boxes will not be able to do).
Linux can not send sound along with video to remote X-servers.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

Rich Alderson <news@alderson.users.panix.com> wrote:
(snip)
 
> News to the folks at XX2247, successors of Mentec.

There seems to be a xx2247.org, but isn't org supposed
to be for non-profit organizations?

-- glen

On 10/6/10 7:26 AM, Bill Gunshannon wrote:
> In article<S_WdnR97l-pXkTHRnZ2dnUVZ_tqdnZ2d@bresnan.com>,
> 	GreyCloud<mist@cumulus.com>  writes:
>> Bill Gunshannon wrote:
>>> In article<Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>,
>>> 	GreyCloud<mist@cumulus.com>  writes:
>>>> Michael Kraemer wrote:
>>>>> In article<4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com>, JF Mezei
>>>>> <jfmezei.spamnot@vaxination.ca>  writes:
>>>>>> Also, if you need another OS to manage a server OS, you might as well
>>>>>> use that other OS as your server OS.
>>>>>>
>>>>> exactly.
>>>>> Moreover, to come up with another of those popular
>>>>> car analogies: what some people suggest here is
>>>>> to sell a car restricted to engine and tank ("server-centric").
>>>>> If you want to actually drive it,
>>>>> you'd have to buy/use a second one,
>>>>> which will give you seats, tires and a driving wheel,
>>>>> so you can hook up the first one.
>>>>> I think most, if not all customers would just
>>>>> buy the second, complete one, even if it
>>>>> has a "weaker" (by whatever measure) engine
>>>>> than the first one.
>>>> You've got a good point.
>>>> Look at cummins diesel plant.  They don't make complete trucks, but they
>>>> do sell their engines to other companies that do.  One would thing that
>>>> HP could sell their engine to others that would complete the product
>>>> offering.
>>>>
>>>> But as I asked JF, why doesn't HP just sell VMS to someone else if they
>>>> continue to hide it from the world on their web site?  Something just
>>>> doesn't add up here.
>>>
>>> Probably because the perciened cost of selling the product line is greater
>>> than any percieved benefit.
>>>
>>
>> That would really depend on what company would want to buy it and how
>> well they can enhance the product to be something desirable for the
>> masses.
>
> That has nothing to do with the cost of selling it.  It's not a bargain-
> baement PC.  Sale of VMS would involve lawyers and contracts and god
> only knows what else.  It could easily (and probably would) cost more
> to transfer VSM than they could hope to find a sucker to pay.
>

That part I hadn't considered.  Yes, I can see a lot of lawyers licking 
their chops if this occurred.

>>            After all, it took someone like Steve Jobs to take a unix os
>> and transform it into what it is now.
>
> Oh please.  Apple still makes up a very small part of the true Unix
> market (remember most of the users of things like iPhones and iPads
> have no idea and couldn't care less that it runs a version of Unix)
> and their love is for Aapple and not Unix.  It's religion, not tech-
> nology.  And then we have Linux which has considerably more market
> share thanApple will ever see.  Jobs is just very good at selling
> snake-oil.  I saw all the presentations fromt he realease of the new
> iPhone and it reminded very strongly of a traveling medicine show.
>

I don't care about the iPhones, iPads, etc.  Just their higher end 
machines and servers that now use Certified UNIX.
In the beginning, it was just a Melon Univeristy that created Mach. 
Jobs people transformed it over time into a full fledged certified UNIX.

The fact that there is a Wiley book called "Advanced Programming in the 
UNIX Environment" that does cover OS X, Solaris, Linux and BSD.

>>                                         I see no reason why the same
>> can't be done with VMS.
>
> "It's the applications  stupid!" (Not attacking, just paraphrasing something
> most people will probably remember. :-)
>
> The amount of time, effort and money it would take it would take to
> catch up now pretty much leaves VMS out of the mainstream for the
> rest of its existence, however long that may be.
>

It could be at this time an opportunity for some company willing to risk 
the venture.  Just like Apple had to do when their OS 9 line was 
faltering and Jobs had to do something to turn the company around.
The fact that DEC, Data General, Apollo, SGI, etc. failed, was largely 
due to top managements failure to analyze the computing trends and were 
incapable of making the necessary changes.

On 10/6/10 7:46 AM, Michael Kraemer wrote:
> In article<S_WdnR97l-pXkTHRnZ2dnUVZ_tqdnZ2d@bresnan.com>, GreyCloud
> <mist@cumulus.com>  writes:
>
>> After all, it took someone like Steve Jobs to take a unix os
>> and transform it into what it is now.  I see no reason why the same
>> can't be done with VMS.
>
> He didn't "transform a unix os".
> He essentially trashed the old OS 9, took a BSD Unix as is,
> stacked a new GUI and a couple of cool apps on top of it,
> and sold it as OS X. Together with the then new G4 and G5 PPCs
> that was an attractive product sold in numbers of 1 ore 2M per year,
> so the costs could be recouped quick enough.

If I recall he took the Mellon Universitys Mach kernel.
Plus jobs already had purchased NextStep, (now known as Cocoa) and 
developed a nice UI.  That's a lot of work.

> There's zero chance to repeat that for VMS because
> all ingredients are missing.
> Certainly you wouldn't like the old VMS be trashed and replaced
> by some Unix (but maybe I'm wrong, given the recent comments on
> VMS 8.4 :-). A new GUI and cool desktop apps are forbidden
> by the authorities here in c.o.v because VMS "is as server OS",
> so out the door with the prospect of selling 1 or 2M units per year
> to recoup costs before entering bankrupcy.
> And, finally, there's no Itanic equivalent in sight for the G4 and
> G5 CPUs.

Erm... IBM now has the G6 and G7 for their servers as I recall from 
digging into their website a bit.  But they are server oriented and eat 
up a lot of power.  IBM, on apples request for lower power consumption 
processors, wouldn't do it... so Apple went to Intel.  I've got an older 
G5 and in some areas does a better job than my newer 2 core Intel.
But, industry moves on and nothing I can do about it.

On 10/6/10 12:05 AM, JF Mezei wrote:
> GreyCloud wrote:
>> Bob Koehler wrote:
>
>> Still tho... HP hides VMS very well.
>
>
> HP doesn't actively hide VMS. They just don't promote it. It is there,
> itn its little corner, visible to those who look for it, but invisible
> to those passing by.

I have already accepted that VMS is pretty much at its dead end.

>
> No amount of bickering will bring VMS back. If I have accepted it and
> moved on, then everyone else should.

I suppose if there is any industry left, and it only takes one company 
willing to take the risk could possibly build a decent UI around VMS... 
but not very likely.

>
> Those lucky enough to still have gainful VMS related employment should
> cherish it.

Oh, I agree there.  Of course those jobs, as another thread has 
indicated, are rapidly disappearing.  From what I can glean from various 
newsfeeds, portable devices are now the in-thing to program for, and 
that market is rapidly expanding.  Being retired from the field for 
about twenty years now, I'm not sure whose servers will handle this 
market the best.

In article <b8ydnV9mbqDpDjPRnZ2dnUVZ_oOdnZ2d@bresnan.com>, GreyCloud
<mist@cumulus.com> writes:
> 
> If I recall he took the Mellon Universitys Mach kernel.
> Plus jobs already had purchased NextStep, (now known as Cocoa) and 
> developed a nice UI.  That's a lot of work.

Of course it is, and expensive too.
But with sales numbers in the millions, 
he could expected to recoup the expenses.
Even back then, Apple probably sold more Macs per year
than VMS boxes have ever been sold in all its existence.
 
> > There's zero chance to repeat that for VMS because
> > all ingredients are missing.
> > Certainly you wouldn't like the old VMS be trashed and replaced
> > by some Unix (but maybe I'm wrong, given the recent comments on
> > VMS 8.4 :-). A new GUI and cool desktop apps are forbidden
> > by the authorities here in c.o.v because VMS "is as server OS",
> > so out the door with the prospect of selling 1 or 2M units per year
> > to recoup costs before entering bankrupcy.
> > And, finally, there's no Itanic equivalent in sight for the G4 and
> > G5 CPUs.
> 
> Erm... IBM now has the G6 and G7 for their servers as I recall from 
> digging into their website a bit.  But they are server oriented and eat 
> up a lot of power.

The "G" is Apple speak, in fact the G4 is a Motorola MPC74xx
and the G5 is an IBM PPC970, a Power4 derivative.
In the meantime, IBM went to Power5, 6 and 7,
which are of course server chips, that's what they were designed for.
But the Power7 doesn't look that big, even with heat sink mounted,
and is rather power efficient (for its 8 cores),
so a lower power version is conceivable.
Rumour has it, that the next Playstation might use it.

>  IBM, on apples request for lower power consumption 
> processors, wouldn't do it... 

Probably Jobs wanted it for free,
and that would have been a bit expensive for Moto/IBM,
given the relatively low volume of Mac sales.
M$ OTOH ordered (and got) a Power CPU for their XBox
(as did Sony and Nintendo for their's),
but those consoles sell at least an order of magnitude more
than Macs.

But that's not the point in the VMS/Apple analogy.
By the time Jobs resurrected the ailing company
(must be around 2000 iirc) their was an upgrade path
for their CPU (G3 => G4,G5). I don't see that for 
Itanic, even the latest and greatest is already behind x86 and P7,
and the next one is years away. 

> so Apple went to Intel.  I've got an older 
> G5 and in some areas does a better job than my newer 2 core Intel.
> But, industry moves on and nothing I can do about it.
 

In article <fPydndoibJppDDPRnZ2dnUVZ_tidnZ2d@bresnan.com>,
	GreyCloud <mist@cumulus.com> writes:
> On 10/6/10 7:26 AM, Bill Gunshannon wrote:
>> 
>> Oh please.  Apple still makes up a very small part of the true Unix
>> market (remember most of the users of things like iPhones and iPads
>> have no idea and couldn't care less that it runs a version of Unix)
>> and their love is for Aapple and not Unix.  It's religion, not tech-
>> nology.  And then we have Linux which has considerably more market
>> share thanApple will ever see.  Jobs is just very good at selling
>> snake-oil.  I saw all the presentations fromt he realease of the new
>> iPhone and it reminded very strongly of a traveling medicine show.
>>
> 
> I don't care about the iPhones, iPads, etc.  Just their higher end 
> machines and servers that now use Certified UNIX.

Why do you place a value on the term "Certified UNIX"?  All that means
is someone was stupid enough to pay someone else a fee for which they 
get nothing in return.


> In the beginning, it was just a Melon Univeristy that created Mach. 

That's Carnegie-Melon University.  And you can see all the importance the
rest of the Unix world placed on the Mach concept.

> Jobs people transformed it over time into a full fledged certified UNIX.

Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
contains a lot of GNU) running on top of it.  The only Apple contribution
is their wierd window manager.

> 
> The fact that there is a Wiley book called "Advanced Programming in the 
> UNIX Environment" that does cover OS X, Solaris, Linux and BSD.

And why not?  There is nothing unique about OS-X in the first place.

> 
>>>                                         I see no reason why the same
>>> can't be done with VMS.
>>
>> "It's the applications  stupid!" (Not attacking, just paraphrasing something
>> most people will probably remember. :-)
>>
>> The amount of time, effort and money it would take it would take to
>> catch up now pretty much leaves VMS out of the mainstream for the
>> rest of its existence, however long that may be.
>>
> 
> It could be at this time an opportunity for some company willing to risk 
> the venture.  

With no chance to ever recoup their investment.

>               Just like Apple had to do when their OS 9 line was 
> faltering and Jobs had to do something to turn the company around.

All he did was take someone elses work.  Put a window manager in front
of it.  And extend the Macintosh religion a few more years.

> The fact that DEC, Data General, Apollo, SGI, etc. failed, was largely 
> due to top managements failure to analyze the computing trends and were 
> incapable of making the necessary changes.

No real argument there.  But you have to add in for a couple of them that
technology pretty much eliminated their niche.  Like SGI whose strength
was graphics until the gamers drove a business that made the average PC
surpass even SGI's graphics capabilities.
 
But in the case of DEC, there is no doubt in anyone's mind that the
real problem lie with management.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <b8ydnV9mbqDpDjPRnZ2dnUVZ_oOdnZ2d@bresnan.com>,
	GreyCloud <mist@cumulus.com> writes:
> On 10/6/10 7:46 AM, Michael Kraemer wrote:
>> In article<S_WdnR97l-pXkTHRnZ2dnUVZ_tqdnZ2d@bresnan.com>, GreyCloud
>> <mist@cumulus.com>  writes:
>>
>>> After all, it took someone like Steve Jobs to take a unix os
>>> and transform it into what it is now.  I see no reason why the same
>>> can't be done with VMS.
>>
>> He didn't "transform a unix os".
>> He essentially trashed the old OS 9, took a BSD Unix as is,
>> stacked a new GUI and a couple of cool apps on top of it,
>> and sold it as OS X. Together with the then new G4 and G5 PPCs
>> that was an attractive product sold in numbers of 1 ore 2M per year,
>> so the costs could be recouped quick enough.
> 
> If I recall he took the Mellon Universitys Mach kernel.

iRight, no development on Apple's part.  Just took an existing "product"
(one, I might mention, that no one else seemed to be showing any interest
in at all) and incorporated it.

> Plus jobs already had purchased NextStep, (now known as Cocoa) and 
> developed a nice UI.  That's a lot of work.

Well, OS-X is not really NextStep (which would probably require paying
royalties to Adobe as it uses Display Postscript) but the GUI is said
to be derived from OpenStep a Open Source clone of NextStep.  But, again,
the only piece of original work on OS-X is the GUI.  A very tiny peice of
the entire package all of which came from Mach and BSD.

> 
>> There's zero chance to repeat that for VMS because
>> all ingredients are missing.
>> Certainly you wouldn't like the old VMS be trashed and replaced
>> by some Unix (but maybe I'm wrong, given the recent comments on
>> VMS 8.4 :-). A new GUI and cool desktop apps are forbidden
>> by the authorities here in c.o.v because VMS "is as server OS",
>> so out the door with the prospect of selling 1 or 2M units per year
>> to recoup costs before entering bankrupcy.
>> And, finally, there's no Itanic equivalent in sight for the G4 and
>> G5 CPUs.
> 
> Erm... IBM now has the G6 and G7 for their servers as I recall from 
> digging into their website a bit.  But they are server oriented and eat 
> up a lot of power.  IBM, on apples request for lower power consumption 
> processors, wouldn't do it... so Apple went to Intel.  I've got an older 
> G5 and in some areas does a better job than my newer 2 core Intel.
> But, industry moves on and nothing I can do about it.
 
IBM must have soemtng as they have recently re-iterated their commitment
to AIX on the Desktop.

bill 

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8n11c$gon$1@lnx107.hrz.tu-darmstadt.de>,
	m.kraemer@gsi.de (Michael Kraemer) writes:
> In article <b8ydnV9mbqDpDjPRnZ2dnUVZ_oOdnZ2d@bresnan.com>, GreyCloud
> <mist@cumulus.com> writes:
>> 
>> If I recall he took the Mellon Universitys Mach kernel.
>> Plus jobs already had purchased NextStep, (now known as Cocoa) and 
>> developed a nice UI.  That's a lot of work.
> 
> Of course it is, 

Mach kernel, written and developed by CMU.
BSD Userland (with GNU additions)

Where exactly is all the work done by Apple?

Oh wait, they did provide their own window manager instead of the dozen
or so already available.

>                   and expensive too.

How is relying on the work of others expensive?
It's not their OS anymore.  It's just another package of Unix components
that are available to anyone who wants them.

> But with sales numbers in the millions, 
> he could expected to recoup the expenses.

Actually, considering how little an investment Apple has in any of that
software and the premium prices paid for it their profits must be sky
high.

> Even back then, Apple probably sold more Macs per year
> than VMS boxes have ever been sold in all its existence.

No argument here, but then, that applies to pretty much everybody.  I'll
bet that in its heyday Tandy had more of their Business Computers (meaning
not the Model I, Model III or COCO) than there were VMS boxes.  But you
also have to take into consideration Apple's mistakes in the IT world.
How many Lisas were sold?  I have only ever seen one once.  It belonged
to a major government contractor who had to provide software for MACs
to the government and, as we all know, development for the MAC was done
on the Lisa.  It had a very short life within my sphere of the country
and eventually just seemed to disappear.

>  
>> > There's zero chance to repeat that for VMS because
>> > all ingredients are missing.
>> > Certainly you wouldn't like the old VMS be trashed and replaced
>> > by some Unix (but maybe I'm wrong, given the recent comments on
>> > VMS 8.4 :-). A new GUI and cool desktop apps are forbidden
>> > by the authorities here in c.o.v because VMS "is as server OS",
>> > so out the door with the prospect of selling 1 or 2M units per year
>> > to recoup costs before entering bankrupcy.
>> > And, finally, there's no Itanic equivalent in sight for the G4 and
>> > G5 CPUs.
>> 
>> Erm... IBM now has the G6 and G7 for their servers as I recall from 
>> digging into their website a bit.  But they are server oriented and eat 
>> up a lot of power.
> 
> The "G" is Apple speak, in fact the G4 is a Motorola MPC74xx
> and the G5 is an IBM PPC970, a Power4 derivative.
> In the meantime, IBM went to Power5, 6 and 7,
> which are of course server chips, that's what they were designed for.
> But the Power7 doesn't look that big, even with heat sink mounted,
> and is rather power efficient (for its 8 cores),
> so a lower power version is conceivable.
> Rumour has it, that the next Playstation might use it.

I thought Playstation was the one using Cell?

> 
>>  IBM, on apples request for lower power consumption 
>> processors, wouldn't do it... 
> 
> Probably Jobs wanted it for free,
> and that would have been a bit expensive for Moto/IBM,
> given the relatively low volume of Mac sales.
> M$ OTOH ordered (and got) a Power CPU for their XBox
> (as did Sony and Nintendo for their's),
> but those consoles sell at least an order of magnitude more
> than Macs.

I imagine game machines outsell everything but the PC.

> 
> But that's not the point in the VMS/Apple analogy.
> By the time Jobs resurrected the ailing company
> (must be around 2000 iirc) their was an upgrade path
> for their CPU (G3 => G4,G5). I don't see that for 
> Itanic, even the latest and greatest is already behind x86 and P7,
> and the next one is years away. 

Again, I agree.  I see no chance that Itanium will ever catch/stay up
with other current cpus and I certainly would not invest in a product
that was tied solely to it.

> 
>> so Apple went to Intel.  I've got an older 
>> G5 and in some areas does a better job than my newer 2 core Intel.
>> But, industry moves on and nothing I can do about it.
>  

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

On 2010-10-07, Rich Alderson <news@alderson.users.panix.com> wrote:
> Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP> writes:
>
>> On 2010-10-06, Rich Alderson <news@alderson.users.panix.com> wrote:
>
>>> News to the folks at XX2247, successors of Mentec.
>
>> What exactly do you mean by Mentec successor ?
>
>> Is this a successor in the legal sense of the word or a unrelated
>> organisation ? Are they resuming Mentec's PDP-11 OS operations ?
>
> Yes, a legal successor.
>
> I thought they would have said something for themselves by now, at least on
> their site.  I shouldn't have said anything, I guess.
>

Thanks for the feedback. Clearly there must still be a market for the PDP-11
environment so it will be interesting to see what services they think are
still viable to offer.

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980's technology to a 21st century world

On 2010-10-07, John Wallace <not@any.price> wrote:
>
> "Bill Gunshannon" <billg999@cs.uofs.edu> wrote in message 
> news:8h3vo2Fv5jU2@mid.individual.net...
>>
>> Now that I would be interested in.  The only PDP-11 Bliss compiler
>> I was aware of was for the PDP-10 and was not available in source
>> anywhere I could find it.  Given a Bliss compiler in source to start
>> with I might try writting a Bliss to C translator myself (in my
>> copious freetime. :-)  Unless, of course it is like the current state
>> of "free" ADA.
>>
>
> "the current state of "free" ADA."
>
> It's Ada. Not ADA. Named after Ada Lovelace.
>
> And what exactly does Bill see as the state of "free" Ada?
>
> What most of the world sees is that companies like Boeing, Airbus, and their 
> suppliers, rely on free (as in open source, as in freely downloadable) Ada 
> (GNAT, the gcc variant which supports Ada) or its commercially supported big 
> brother from AdaCore, GNAT Pro. Same gcc source code whichever one you pick 
> (give or take a few months bugfixing, as their GPL one is released once or 
> twice a year, I forget, whereas the commercial one presumably gets bugfixes 
> earlier).
>

To add to John's comments, I should mention also that there are 2 distinct
sources for the freely available version of GNAT.

The original source is the version of GNAT supplied by the FSF as part of
the mainline GCC tree. This version has no restrictions on what software
can be developed with it and like the GPL version below was created by and
is maintained by ACT.

ACT also maintain another free branch available on their website and this
version is released for use with GPL compatible projects only. What you
apparently get with this version over the FSF version is a more integrated
and tested release.

My personal choice is for the FSF version because of it's freedom and I have
yet to experience any real quality problems with it. I also use Ada within
RTEMS and RTEMS is developed against the FSF version of GCC.

> GNAT even has VMS as a commercially supported platform (probably not 
> unrelated to DEC's decision not to develop an Ada 9x compiler).
>

There also used to be a freely available public release of GNAT for VMS,
but this was back in the GCC 2.8.x days.

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980's technology to a 21st century world

In article <8h8hhaF434U3@mid.individual.net>, billg999@cs.uofs.edu (Bill
Gunshannon) writes:
> 
> Actually, considering how little an investment Apple has in any of that
> software and the premium prices paid for it their profits must be sky
> high.

even if they reused someone else's kernel,
a GUI is lot of work, as is testing and integration.
And the old OS 9 stuff still had to work too.
 
> 
> I thought Playstation was the one using Cell?
> 

The current one, yes, but Cells are a nightmare
to program for, which is probably why IBM/Sony don't follow
this sidebranch any longer
(unlike certain other companies who keep riding a dead horse).
A stripped-down multicore Power7 would certainly be more attractive
than those Cell SPEs.

In article <8h8gd0F434U1@mid.individual.net>, billg999@cs.uofs.edu (Bill Gunshannon) writes:

> Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
> contains a lot of GNU) running on top of it.  The only Apple contribution
> is their wierd window manager.
 
[...]
 
> And why not?  There is nothing unique about OS-X in the first place.

   That GUI is a lot more than just a "window manager".  And although
   you find it wierd, it's the best GUI over a UNIX that I've ever
   seen.

   You wanted an old Sun, maybe?

In article <oOlLv1mVX$cE@eisner.encompasserve.org>,
	koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:
> In article <8h8gd0F434U1@mid.individual.net>, billg999@cs.uofs.edu (Bill Gunshannon) writes:
> 
>> Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
>> contains a lot of GNU) running on top of it.  The only Apple contribution
>> is their wierd window manager.
>  
> [...]
>  
>> And why not?  There is nothing unique about OS-X in the first place.
> 
>    That GUI is a lot more than just a "window manager".  And although
>    you find it wierd, it's the best GUI over a UNIX that I've ever
>    seen.

Sounds like more Mac religion to me.

> 
>    You wanted an old Sun, maybe?
 
No, although at the time it was pretty good.  Much better than the
one-button-mouse GUI on the Mac of the same era.

There are dozens of window mannagers to choose from. There have even
been some done that mimiced the look and feel of Windows (they didn't
last long as they really brought nothing new, unique or particularly
useful to the table.)  Too the best of my knowledge, there has never
even been an attempt to write one that mimiced the look and feel of
the Mac GUI.  Now what do you think that means?

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

Michael Kraemer wrote:

> By the time Jobs resurrected the ailing company
> (must be around 2000 iirc) their was an upgrade path
> for their CPU (G3 => G4,G5). I don't see that for 
> Itanic, even the latest and greatest is already behind x86 and P7,
> and the next one is years away. 

In fairness, Intel/HP have published a roadmap for that IA64 thing which
includes 2 releases beyond the current one. Of course, Alpha also had 2
releases on its roadmap on June 24 2001.

Apple switched to the 8086 because it couldn't get laptop CPUs quickly
enough and its laptops would lag the 8086 based ones. IBM was busy
catering to the game consoles.


Alpha was killed at a time when it still had a fair amount of
performance edge. Even the much delayed EV7 packed enough punch.

However, the slow development of IA64 has now resulted in the latest
chip not having any punch at all when it was released because
competitors were already ahead of it.

When HP/Intel announce the EOL of IA64, it will be because customers
have been telling them they prefer to use industry standard servers and
that IA64 doesn't bring any performance advantage to them. Such a
decision to kill off IA64 won't generate much anger from customers.

Intel will likely announce some delay with Poulson (that is the next
IA64, right ?). Then, after the next 8086 is released, the performance
gap between 8086 and IA64 will allow HP/Intel to announce that IA64's
relative performance is on a downward curve and that it makes sense to
move to the 8086 for all servers.

Bill Gunshannon wrote:

> Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
> contains a lot of GNU) running on top of it.  The only Apple contribution
> is their wierd window manager.

Apple did not just "use" Mach. It fixed it. There were good reason why
it had not been widely adopted in the past. Apple fixed those reasons.
(I rememer reading an article about this, but the technical aspect elude
me at the moment).


> And why not?  There is nothing unique about OS-X in the first place.

Devout members of the Church of Apple would disagree. OS-X sports some
very significant attributes important to them, the most important of
which is the Apple logo. :-)

You greatly underestimate OS-X if you think Apple only added a weird
window manager. There are a large number of frameworks to help you build
apps. It is a very rich programming environment.


> With no chance to ever recoup their investment.

I would disagree. Both Linux and Apple were able to enter the market
"late in the game" and displace Windows. With sufficient investment, VMS
could carve a healthy piece of the pie for itself. EVen 1% of the market
would be a huge boost for VMS.

However, for VMS to be revived today, it would only work with a total
commitment. No half measures or token advertising will do.


> All he did was take someone elses work.  Put a window manager in front
> of it.  And extend the Macintosh religion a few more years.

Consider the number of Apple places of Worship that have been built in
recent years, including a spectacular one in Shanghai. I'd say the Apple
religion is alive and doing rather well. Wall Street would tend to agree
too.

In one of the keynotes, Jobs had mentioned the number of people who
visit those places of worship every day and the number was astounding.

And it is ironic to hear such "religious" talk here. Consider that in
the 1980s, there was a devout Church of Digital, where every DEC
customer was expected to attend at least one DECUS event in his
lifetime.  And instead of walking around the Ka'bah and throwing stones
at it, DECUS attendees went around the demo room and tested DEC's gear
and spoke to engineers. The rituals were strong.


The loyalty to Digital and its products was very high. and Deccies had
immense loyalty to Digital and its products. There were religious wars
between VMS and Unix.

glen herrmannsfeldt <gah@ugcs.caltech.edu> writes:

> Rich Alderson <news@alderson.users.panix.com> wrote:
> (snip)

>> News to the folks at XX2247, successors of Mentec.

> There seems to be a xx2247.org, but isn't org supposed
> to be for non-profit organizations?

Not for a very very long time now.

There also used to a division among IP addresses among class A (1.n.n.n to
127.n.n.n), class B (128.n.n.n to 191.n.n.n) and class C (192.n.n.n to
254.n.n.n).  You do know that that changed as well, right?

;-)

-- 
Rich Alderson                                   news@alderson.users.panix.com
    the russet leaves of an autumn oak/inspire once again the failed poet/
    to take up his pen/and essay to place his meagre words upon the page...

"JF Mezei" <jfmezei.spamnot@vaxination.ca> wrote in message 
news:4caf85fa$0$31638$c3e8da3$c14f6927@news.astraweb.com...
> Michael Kraemer wrote:
>

> However, the slow development of IA64 has now resulted in the latest
> chip not having any punch at all when it was released because
> competitors were already ahead of it.
>

CPU's in general are losing their "punch" and hitting a wall with single 
threaded performance.  Which is why what you see is a proliferation of 
multiple cores and a re-focus on memory performance.  IA64 isn't nearly as 
bad as you think it is.  The latest blade servers are *very* good.

Lots of cores only solve some problems - and create others - but its what 
the chip designers are doing to try to make up for the stagnation in single 
thread performance.

On Fri, 2010-10-08 at 17:23 -0400, JF Mezei wrote:
> I would disagree. Both Linux and Apple were able to enter the market
> "late in the game" and displace Windows. With sufficient investment,
> VMS could carve a healthy piece of the pie for itself. EVen 1% of the
> market would be a huge boost for VMS.
>=20
> However, for VMS to be revived today, it would only work with a total
> commitment. No half measures or token advertising will do.=20

I would seriously consider a X11 makeover to be an essential part of
that revitalisation process.=20

That and ports to other architectures would be nice. The more I use VMS
the more I'm liking it.=20
--=20
Tactical Nuclear Kittens

On 10/6/2010 1:42 AM, GreyCloud wrote:
> Michael Kraemer wrote:
>> In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>, GreyCloud
>> <mist@cumulus.com> writes:
>>
>>> But as I asked JF, why doesn't HP just sell VMS to someone else
>>
>> Why should they?
>> They don't have a problem with it.
>> Contrary to common belief here the world doesn't have problems
>> without it.
>> So why change the status quo?
>
> I thought HP as a company was in business to sell products.
> Either advertise it or sell it off.
>
>>
>>> if they continue to hide it from the world on their web site?
>>
>> Do they?
>> So far you would find all VMS related info on their respective pages.
>> Of course not on the front pages, no wonder
>> since 99+x% of HP customers are interested in PCs and ink
>> rather than nerd stuff such as VMS.
>
> Believe me, it isn't out in the open at HPs website... but their PCs
> being MS OS types are right there in your face at their front page.
>
> It would be like IBM not advertising their OS... which they do and the
> world knows it.


My experience has been that it's almost impossible to find anything on 
the HP web site using HP's search engine..  I use Google and specify 
"site=HP.COM" as part of my search string.

On 10/6/2010 1:44 AM, GreyCloud wrote:
> Bill Gunshannon wrote:
>> In article <Hv6dnbG-gp17LDfRnZ2dnUVZ_rudnZ2d@bresnan.com>,
>> GreyCloud <mist@cumulus.com> writes:
>>> Michael Kraemer wrote:
>>>> In article <4ca98dbe$0$29158$c3e8da3$88b277c5@news.astraweb.com>, JF
>>>> Mezei
>>>> <jfmezei.spamnot@vaxination.ca> writes:
>>>>> Also, if you need another OS to manage a server OS, you might as well
>>>>> use that other OS as your server OS.
>>>>>
>>>> exactly.
>>>> Moreover, to come up with another of those popular
>>>> car analogies: what some people suggest here is
>>>> to sell a car restricted to engine and tank ("server-centric").
>>>> If you want to actually drive it,
>>>> you'd have to buy/use a second one,
>>>> which will give you seats, tires and a driving wheel,
>>>> so you can hook up the first one.
>>>> I think most, if not all customers would just
>>>> buy the second, complete one, even if it has a "weaker" (by whatever
>>>> measure) engine
>>>> than the first one.
>>> You've got a good point.
>>> Look at cummins diesel plant. They don't make complete trucks, but
>>> they do sell their engines to other companies that do. One would
>>> thing that HP could sell their engine to others that would complete
>>> the product offering.
>>>
>>> But as I asked JF, why doesn't HP just sell VMS to someone else if
>>> they continue to hide it from the world on their web site? Something
>>> just doesn't add up here.
>>
>> Probably because the perciened cost of selling the product line is
>> greater
>> than any percieved benefit.
>>
>
> That would really depend on what company would want to buy it and how
> well they can enhance the product to be something desirable for the
> masses. After all, it took someone like Steve Jobs to take a unix os and
> transform it into what it is now. I see no reason why the same can't be
> done with VMS.

Who is going to be cast as Steve Jobs???????

Rich Alderson <news@alderson.users.panix.com> wrote:
(snip, I wrote)
 
>> There seems to be a xx2247.org, but isn't org supposed
>> to be for non-profit organizations?
 
> Not for a very very long time now.

If you mean that there is no enforcement for the distinction,
then I completely agree.  Now, I suppose some might use the .org
domain when the matching .com was already taken, but that doesn't
seem to be true here.

Since many people assume .com, without always reading the URL,
I don't see why anyone with a profit-making company would register
just the .org. 
 
> There also used to a division among IP addresses among class A (1.n.n.n to
> 127.n.n.n), class B (128.n.n.n to 191.n.n.n) and class C (192.n.n.n to
> 254.n.n.n).  You do know that that changed as well, right?

Yes, I know about that one.

I thought class C only went up to 223, do leave room for future classes.
Possibly that was added later, though.

-- glen

> My experience has been that it's almost impossible to find anything on 
> the HP web site using HP's search engine..  I use Google and specify 
> "site=HP.COM" as part of my search string.

That goes for many sites. Don't you love it when you type in a search 
string on a site and it comes back with a slew of pages (most of them 
trying to sell something), none of which contain the search string.

Of course, the same thing happens occasionally on Google, too. Yesterday 
I was searching for specs on an LCD monitor and several of the top 
results were red herrings - i.e., when opened, the search string was 
nowhere to be found. OK, I figured, the content must have changed since 
Google crawled the page. So I opened the cached version, but the search 
string wasn't in it, either.

You wonder why Google does that. Is it just the sheer technical 
difficulty of displaying pages containing a particular search string, or 
do they take money to put irrelevant pages at the top of the results?

On 2010-10-08 23:23, JF Mezei wrote:
> Bill Gunshannon wrote:
>
>> Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
>> contains a lot of GNU) running on top of it.  The only Apple contribution
>> is their wierd window manager.
>
> Apple did not just "use" Mach. It fixed it. There were good reason why
> it had not been widely adopted in the past. Apple fixed those reasons.
> (I rememer reading an article about this, but the technical aspect elude
> me at the moment).

Well, technically, OS-X isn't much Mach at all. It's mostly NextSTEP 
with FreeBSD/NetBSD thrown in. NextSTEP in turn was based on Mach, 
people say, but that is in a very limited way, as NextSTEP was also BSD 
based, with just a few cherry-picked Mach:isms thrown in.
A serious Mach lover would not recognize himself in OS-X.

I know that everyone says that OS-X is Mach based, but I'd like anyone 
to show me where. The Mach kernel was based around a few simple 
concepts. It's a microkernel, with an extensive IPC having features and 
possibilities not existing in more classical Unixes. Neither of these 
two features have been adopted by anything else.
The third part is the VM system, which you might argue all modern BSD 
based systems use, which was adopted from Mach.

So, no. OS-X is definitely not the Mach kernel with the BSD Userland. 
It's mostly BSD kernel with BSD userland. If you can find anything of 
Mach in there, please let me know.

There is quite a lot of information in Wikipedia about Mach, OS-X, 
NextSTEP and XNU, for anyone to read if they are interested.

That said, someone else claimed that Linux was bigger than OS-X. That 
don't seem to be true. OS-X as a Unix system overshadows all other Unix 
variants combined, when it comes to deployed systems (yes, add all 
Linux, Solaris, xBSD, OSF/1, Ultrix, whatever together, and it's fewer 
systems than the OS-X ones...)

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

On 2010-10-08 14:23, Bill Gunshannon wrote:
> In article<b8ydnV9mbqDpDjPRnZ2dnUVZ_oOdnZ2d@bresnan.com>,
> 	GreyCloud<mist@cumulus.com>  writes:
>> On 10/6/10 7:46 AM, Michael Kraemer wrote:
>>> In article<S_WdnR97l-pXkTHRnZ2dnUVZ_tqdnZ2d@bresnan.com>, GreyCloud
>>> <mist@cumulus.com>   writes:
>>>
>>>> After all, it took someone like Steve Jobs to take a unix os
>>>> and transform it into what it is now.  I see no reason why the same
>>>> can't be done with VMS.
>>>
>>> He didn't "transform a unix os".
>>> He essentially trashed the old OS 9, took a BSD Unix as is,
>>> stacked a new GUI and a couple of cool apps on top of it,
>>> and sold it as OS X. Together with the then new G4 and G5 PPCs
>>> that was an attractive product sold in numbers of 1 ore 2M per year,
>>> so the costs could be recouped quick enough.
>>
>> If I recall he took the Mellon Universitys Mach kernel.
>
> iRight, no development on Apple's part.  Just took an existing "product"
> (one, I might mention, that no one else seemed to be showing any interest
> in at all) and incorporated it.
>
>> Plus jobs already had purchased NextStep, (now known as Cocoa) and
>> developed a nice UI.  That's a lot of work.
>
> Well, OS-X is not really NextStep (which would probably require paying
> royalties to Adobe as it uses Display Postscript) but the GUI is said
> to be derived from OpenStep a Open Source clone of NextStep.  But, again,
> the only piece of original work on OS-X is the GUI.  A very tiny peice of
> the entire package all of which came from Mach and BSD.

Well. OpenStep was developed by Next, and was not a "clone". Instead it 
was a stripped down verison of NextStep done in cooperation with Sun. 
The point being that any Mach dependecies be removed, and the resulting 
software should be possible to run on other (all?) Unix dialects.

OpenSTEP still used Display Postscript. And Apple bought Next, getting 
both NextStep and OpenStep by the way, so they already have the rights.

Cocoa is based on OpenStep, but OS-X is no longer OpenStep compliant.

(See http://en.wikipedia.org/wiki/OpenStep)

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

On 2010-10-04 21:31, Bill Gunshannon wrote:
> In article<i8d3ae$1ak$1@usenet01.boi.hp.com>,
> 	"FredK"<fred.nospam@dec.com>  writes:
>>
>>  From an OS perspective VMS is safer than UNIX-like OSes for many aspects of
>> operation because of roots in it's design - part of which was language and
>> calling conventions.  VMS would never have chosen "malloc" as a memory
>> allocation API.  Or depended on zero termination for strings.
>
> When are we going to put this myth to sleep.  Unix didn't invent null
> terminated strings.  DEC was perfectly happy with them as well.  All
> of the non-Unix PDP-11 OSes had them and no matter how hard people here
> try to deny it, so did VMS, from the very beginning (and, I'll bet it's
> still there in MACRO for anyone who wants to use it!)

Bill, you should make a difference between a program having the 
possibility to generate them, and the language and system convention of 
actually forcing you to use them. :-)

I can create a nul terminated string in any language, on any system. 
What does that prove? Nothing. I can just as well create a CR-terminated 
string in any language, on any system. Does that prove anything more? No.

The problem with Unix is that system calls takes nul-terminated strings 
as parameters. And yes, most library calls do, too. It's a philosophical 
concept. The fact that you can create nul-terminated strings in your 
program is not a problem. The fact that you *have* to use nul-terminated 
strings are...

No system call I know of in RSX accepts a nul-terminated string as an 
argument. They all take a pointer and a length argument. As far as I 
know, the same is true in VMS. I'm almost sure the same is also true in 
RSTS/E, leaving only RT-11 as being unsafe. However, RT-11 don't have 
any memory protection either, so it's not a safe system by any standards 
to start with...

	Johnny



-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

On 2010-10-05 15:33, Bill Gunshannon wrote:
> In article<i8dak6$5h9$1@usenet01.boi.hp.com>,
> 	"FredK"<fred.nospam@dec.com>  writes:
>>
>> "Bill Gunshannon"<billg999@cs.uofs.edu>  wrote in message
>> news:8guoccF6t8U1@mid.individual.net...
>>> In article<i8d3ae$1ak$1@usenet01.boi.hp.com>,
>>> "FredK"<fred.nospam@dec.com>  writes:
>>>>
>>>>  From an OS perspective VMS is safer than UNIX-like OSes for many aspects
>>>> of
>>>> operation because of roots in it's design - part of which was language
>>>> and
>>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>>> allocation API.  Or depended on zero termination for strings.
>>>
>>> When are we going to put this myth to sleep.  Unix didn't invent null
>>> terminated strings.  DEC was perfectly happy with them as well.  All
>>> of the non-Unix PDP-11 OSes had them and no matter how hard people here
>>> try to deny it, so did VMS, from the very beginning (and, I'll bet it's
>>> still there in MACRO for anyone who wants to use it!)
>>>
>>
>> Bill, I didn't mean to imply UNIX or C "invented" zwro terminated strings.
>
> Not singling you out.  It is just a common thread here where people
> seem to ignore the fact that Unix/C merely picked up on a method that
> was pretty much standard accross the industry.

I don't agree with you. And you are confusing system calls with 
libraries as well, I suspect.

>> C made it the language standard,
>
> De facto made it a standard.  To the best of my knowledge C does not have
> a string data type at all.

True. It only have pointers and integers of different sizes. However, C 
also have pointer arithmetic and memory dereferencing, which means that 
you can do anything unsafe you could ever imagine. :-) Even without a 
string type. And maybe that was part of the problem. If C had had a 
proper string type, maybe a lot of problems would have been avoided.

>>                                    UNIX enshrined it in it's interfaces.
>
> You mean like the example I posted for VMS?  Use made it what it is and
> people are very unlikely to make major changes that break lots of already
> written software.  That is what caused all the two year dates more than a
> decade after punched-cards went away and keeps time_t too small and even
> keeps IPV6 out of the mainstream (I first used IPV6 more than 10 years
> ago and we actually had (have?) an address allocation).

I don't know the .TTYOUT library function that you used. Never seen it. 
Looks very much like RT-11 to me. :-)

Anyway, I bet that library is just as unsafe as any Unix code, yes. But 
that library is rather in the face of VMS conventions, where you 
actually have a string type defined, which should even be used if you 
write in MACRO-32. And all system calls, as well as standard libraries 
will expect that string type, and not your nul-terminated string, so you 
will have to do the conversion every time you plan to use any standard 
system functions or libraries.
And the standard string type in VMS have a pointer and a length field. 
You should know this... :-)

>> Despite .ASCIZ - find it used as part of the calling standard.
>
> The fact that it is still included as a part of the language tells me
> is is, in fact, a "part of the calling standar" on VMS.  Just because
> people choose to use something better doesn't remove it from the
> standard.

No. The calling standard in VMS is well defined (in contradiction to 
Unix systems, where it is not). Strings is their own type. Just because 
you decide to use nul-terminated strings, and the fact that there are 
libraries that accept these, don't make them the calling standard.

I suspect that in the name of Unix/Posix compatibility, nul terminated 
string have probably spread in the VMS world since I last played VMS 
though, so maybe they are officially a part of the calling standard in 
VMS nowadays. But I very much doubt they show up anywhere in the older 
core of VMS.

>>                                                                   It was
>> stupidity on a grand scale.
>
> Hindsight is always 20/20.  It served a purpose and worked very well for
> a long time.  It is still usable and perfectly safe depending on the
> programmer.   And, in some environments it is perfectly safe regardless.
> I would foresee no problem using .ASCIZ cwunder RT-11, RSTS, RSX or under
> VMS.  Of course, ~i have no problem with using it under Unix, but that's
> me.  :-)

Using nul-terminated string is a potential disaster under any OS. No 
system call under RSX accepts them, and very few library functions will 
have anything to do with them.
In RSX though, there is no system wide calling conventions. You have the 
FORTRAN calling convention, using R5 as an argument pointer. You have 
BASIC+2, which use another convention, you have PDP-11 C, which use yet 
another. And they all represent strings in their own ways...
Which also means that you need to convert strings whenever you mix 
things, or when you do system calls.

I try to stay away from nul-terminated string for most things, and only 
use them when I'm lazy, and for very restricted and specific things.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

On 2010-10-05 03:32, Michael Kraemer wrote:
> In article<8guoccF6t8U1@mid.individual.net>, billg999@cs.uofs.edu (Bill
> Gunshannon) writes:
>>
>> When are we going to put this myth to sleep.  Unix didn't invent null
>> terminated strings.  DEC was perfectly happy with them as well.  All
>> of the non-Unix PDP-11 OSes had them
>
> exactly.
> The first time I encountered them (in a former life)
> was on a PDP-11 under RSX-11-something.
> It seemed to be the standard way to store varying length strings, IIRC.
> Quite efficient IMHO for an architecture which maxes out at 16 bits,
> where a 2-byte length prefix might be considered as a waste of resources
> compared to a 1-byte trailing null byte.

Huh?
I've been using RSX for over 20 years, and am still actively using it. 
Nul-terminated strings are hardly the norm anywhere in the system.

Yes, you can (as always) create nul-terminated strings, and yes, 
MACRO-11 even have an operation for it.
However, no system call take that as an argument, and I can think of 
about 1 library function that plays around with nul-terminated strings 
($EDMSG).

What some people might have done, when writing in assembler, to maximize 
the memory usage is another thing, though. But don't point at RSX for that.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

On 2010-10-04 22:35, JF Mezei wrote:
> FredK wrote:
>
>> calling conventions.  VMS would never have chosen "malloc" as a memory
>> allocation API.
>
> I ask this naively. What is wrong with malloc ?  You give it a size, it
> returns a pointer to a memory block allocated to you of that size.

I don't see anything wrong with that, and I suspect the VMS library call 
to allocate more memory is pretty similar in its interface.
I know the RSX library call is...

malloc is in it self not a problem.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

On 2010-10-04 23:01, FredK wrote:
> "JF Mezei"<jfmezei.spamnot@vaxination.ca>  wrote in message
> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>> FredK wrote:
>>
>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>> allocation API.
>>
>> I ask this naively. What is wrong with malloc ?  You give it a size, it
>> returns a pointer to a memory block allocated to you of that size.
>>
>
> It also doesn't return an error code.

Yes it do.
If malloc fails, it returns a NULL pointer, and sets errno. Very much in 
the same tradition as all Unix library functions.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

On 2010-10-05 00:32, FredK wrote:
> "Bob Eager"<rde42@spamcop.net>  wrote in message
> news:8gv1rnFefcU5@mid.individual.net...
>> On Mon, 04 Oct 2010 17:01:02 -0400, FredK wrote:
>>
>>> "JF Mezei"<jfmezei.spamnot@vaxination.ca>  wrote in message
>>> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>>>> FredK wrote:
>>>>
>>>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>>>> allocation API.
>>>>
>>>> I ask this naively. What is wrong with malloc ?  You give it a size, it
>>>> returns a pointer to a memory block allocated to you of that size.
>>>>
>>>>
>>> It also doesn't return an error code.
>>
>> If it complies with the C standard, it does.
>>
>
> foo = malloc(size);
>
> The only "error" return is NULL.   Returning NULL provides me with zero
> information that can be either used to correct the problem, or instruct the
> user how to correct the problem - other than guessing.  So to be more long
> winded, if there can be more than one possible error failure, a function
> should return a seperate error code.
>
> malloc was simply an example of the generic aversion to error handling,
> followed by kludges to provide errors.

All Unix functions do the same thing. Look at fopen() if you want. It 
returns a file handle, or a NULL pointer.

Have the errno variable totally sailed you by?

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

On 2010-10-05 15:47, Bill Gunshannon wrote:
> In article<i8dikb$9tj$1@usenet01.boi.hp.com>,
> 	"FredK"<fred.nospam@dec.com>  writes:
>>
>> "Bob Eager"<rde42@spamcop.net>  wrote in message
>> news:8gv1rnFefcU5@mid.individual.net...
>>> On Mon, 04 Oct 2010 17:01:02 -0400, FredK wrote:
>>>
>>>> "JF Mezei"<jfmezei.spamnot@vaxination.ca>  wrote in message
>>>> news:4caa3a95$0$23811$c3e8da3$9deca2c3@news.astraweb.com...
>>>>> FredK wrote:
>>>>>
>>>>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>>>>> allocation API.
>>>>>
>>>>> I ask this naively. What is wrong with malloc ?  You give it a size, it
>>>>> returns a pointer to a memory block allocated to you of that size.
>>>>>
>>>>>
>>>> It also doesn't return an error code.
>>>
>>> If it complies with the C standard, it does.
>>>
>>
>> foo = malloc(size);
>>
>> The only "error" return is NULL.
>
> Wrong.  The NULL is not an error return it is how much memory was allocated.

No. NULL return is the error return value. On successful call, a pointer 
to the allocated memory is return. On error, NULL is returned.

>>                                    Returning NULL provides me with zero
>> information that can be either used to correct the problem, or instruct the
>> user how to correct the problem - other than guessing.
>
> Unless, you decide to read tha man page which tells you to check errno.

Indeed.

>>                                                           So to be more long
>> winded, if there can be more than one possible error failure, a function
>> should return a seperate error code.
>
> There is only one possible failure of malloc().  No memory to allocate.
> And it does return a separate error code even though there is only one.

Right.

> I can think of one possible reason why malloc() would return NULL and
> have it not be an error.  if you asked for NULL bytes.  Otherwise the
> only cause for a return of NULL would be not enough memory available to
> meet the request and that is handled perfectly well as documented.

I have not tried this, but as far as I read the man-page, I would 
suspect that calling malloc(0) should return a pointer to a memory 
address, but one where you don't have any bytes to play with. Like a 
void pointer.

It's like a malloc(0) should always succeed, since you can always 
allocate 0 bytes of memory, those you will never run out of.

It would be weird to get a ENOMEM when you did a malloc(0).

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

In article <i8pms6$v0t$1@iltempo.update.uu.se>,
	Johnny Billquist <bqt@softjar.se> writes:
> On 2010-10-08 23:23, JF Mezei wrote:
>> Bill Gunshannon wrote:
>>
>>> Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
>>> contains a lot of GNU) running on top of it.  The only Apple contribution
>>> is their wierd window manager.
>>
>> Apple did not just "use" Mach. It fixed it. There were good reason why
>> it had not been widely adopted in the past. Apple fixed those reasons.
>> (I rememer reading an article about this, but the technical aspect elude
>> me at the moment).
> 
> Well, technically, OS-X isn't much Mach at all. It's mostly NextSTEP 
> with FreeBSD/NetBSD thrown in. NextSTEP in turn was based on Mach, 
> people say, but that is in a very limited way, as NextSTEP was also BSD 
> based, with just a few cherry-picked Mach:isms thrown in.
> A serious Mach lover would not recognize himself in OS-X.
> 
> I know that everyone says that OS-X is Mach based, but I'd like anyone 
> to show me where. The Mach kernel was based around a few simple 
> concepts. It's a microkernel, with an extensive IPC having features and 
> possibilities not existing in more classical Unixes. Neither of these 
> two features have been adopted by anything else.
> The third part is the VM system, which you might argue all modern BSD 
> based systems use, which was adopted from Mach.
> 
> So, no. OS-X is definitely not the Mach kernel with the BSD Userland. 

OK, I'll concede that although it really appears to depend on where you
read.

> It's mostly BSD kernel with BSD userland. 

Which would make it even more just another distro of Unix.

>                                            If you can find anything of 
> Mach in there, please let me know.
> 
> There is quite a lot of information in Wikipedia about Mach, OS-X, 
> NextSTEP and XNU, for anyone to read if they are interested.

I am  not inclined to tear it appart to see what's at the guts.  It's
a religion anyway and I am not likely to conbince anyone anyway.

> 
> That said, someone else claimed that Linux was bigger than OS-X. That 
> don't seem to be true. OS-X as a Unix system overshadows all other Unix 
> variants combined, when it comes to deployed systems (yes, add all 
> Linux, Solaris, xBSD, OSF/1, Ultrix, whatever together, and it's fewer 
> systems than the OS-X ones...)

Well, I assume you are getting these numbers by counting iPhones and
other devices with embeded OS-X.  If that's the case yopu really need
to count all the systems with embedded versions of the other OSes as
well.

How many Netgear switches have been sold world-wide?
How many Linksys Access Points?
How many Linksys/Cisco NSS boxes?

All of them have Linux under the hood.  And there are hundreds more
each having sold thousands of units.

The original argument was about desktop usage as opposed to server
usage and wasn't really about embeded systems at all.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

On 2010-10-05 19:56, Bill Gunshannon wrote:
> In article<i8dq16$d9r$1@usenet01.boi.hp.com>,
> 	"FredK"<fred.nospam@dec.com>  writes:
>>
>> Thanks.  Essentially because certain routines didn't include an error
>> return, they created the errno hack.  Quite a hack... a global symbol.
>
> Matter of opinion.  It has served Unix programmers well for a lot longer
> than VMS has even been around.

The same solution exists in most OSes. In VMS, it is instead returned in 
R0 from system calls, which is better since it can easily be made thread 
safe. But it is still just a value set somewhere to tell whether the 
call succeeded or not.

In RSX, the value is stored in $DSW, which serves exactly the same 
purpose as errno in Unix.

>> I can defend things like C not having a general string type if C is viewed
>> as a assembler replacement as opposed to a high-level language (although a
>> counted string would make more sense in general than null terminated).
>
> If you want a counted string, create it.  That's one of the beauties of
> C.  The basic types as might be defined at the hardware level already
> exist and you are free to create any other user defined type you feel
> necessary.  Maker a string what you want a string to be.  I have worked
> with counted strings (ala UCSD Pascal) I saw no real advantage to them.
> And one very big disadvantage.  They come with a very finite fixed maximum
> length.  If nothing else, that either keeps them very short or makes them
> very non-portable.

One of the problems with this in Unix is that you need to convert it 
back to a nul-terminated string for system calls and library calls. 
Which is why it is a problem in Unix. Unix does not appreciate counted 
strings.

As for your experience with Pascal, I don't see the relevance. There is 
no need to place the string length in the first byte of the string, and 
limit string lengths to 255 bytes. That is just stupid, and it is not 
the only alternative.

>> The language decision led to OS API decisions.
>
> How?  The language does not define a "string" type.  That is actually
> what is in the OS API.  And there have been numerous oportunities to
> change it.  No one showed any interest. (Or have we all forgotten things
> line "Safe C" which was created back in the 70's and as a commercial
> product, didn't fly.)

You cannot change the system calls on a whim. Everything will break.
And that is the root of Unix problem. The API is not well though out 
from a safety point of view, and we're stuck with it.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

In article <i8pp89$vo4$1@iltempo.update.uu.se>,
	Johnny Billquist <bqt@softjar.se> writes:
> On 2010-10-05 15:33, Bill Gunshannon wrote:
>> In article<i8dak6$5h9$1@usenet01.boi.hp.com>,
>> 	"FredK"<fred.nospam@dec.com>  writes:
>>>
>>> "Bill Gunshannon"<billg999@cs.uofs.edu>  wrote in message
>>> news:8guoccF6t8U1@mid.individual.net...
>>>> In article<i8d3ae$1ak$1@usenet01.boi.hp.com>,
>>>> "FredK"<fred.nospam@dec.com>  writes:
>>>>>
>>>>>  From an OS perspective VMS is safer than UNIX-like OSes for many aspects
>>>>> of
>>>>> operation because of roots in it's design - part of which was language
>>>>> and
>>>>> calling conventions.  VMS would never have chosen "malloc" as a memory
>>>>> allocation API.  Or depended on zero termination for strings.
>>>>
>>>> When are we going to put this myth to sleep.  Unix didn't invent null
>>>> terminated strings.  DEC was perfectly happy with them as well.  All
>>>> of the non-Unix PDP-11 OSes had them and no matter how hard people here
>>>> try to deny it, so did VMS, from the very beginning (and, I'll bet it's
>>>> still there in MACRO for anyone who wants to use it!)
>>>>
>>>
>>> Bill, I didn't mean to imply UNIX or C "invented" zwro terminated strings.
>>
>> Not singling you out.  It is just a common thread here where people
>> seem to ignore the fact that Unix/C merely picked up on a method that
>> was pretty much standard accross the industry.
> 
> I don't agree with you. And you are confusing system calls with 
> libraries as well, I suspect.

Actually, I was basing my statement on the fact that all of the PDP-11
OSes that pre-existed Unix had the .ASCIZ directive as a builtin part
of their MACRO assemblers.  Kind of demonstrates that null-terminated
strings predate Unix and were common practice in the IT world.  And,
VMS continued this practice in the first VMS MACRO assembler meaning
its engineers obviously didn't see any threat from it at that time.

But wait, it gets better.

> 
>>> C made it the language standard,
>>
>> De facto made it a standard.  To the best of my knowledge C does not have
>> a string data type at all.
> 
> True. It only have pointers and integers of different sizes. However, C 
> also have pointer arithmetic and memory dereferencing, which means that 
> you can do anything unsafe you could ever imagine. :-) Even without a 
> string type. And maybe that was part of the problem. If C had had a 
> proper string type, maybe a lot of problems would have been avoided.

All of this had been fixed in the early 70's.  Ever hear of "Safe-C"?
it was met with a resounding silence and the company just went away.

> 
>>>                                    UNIX enshrined it in it's interfaces.
>>
>> You mean like the example I posted for VMS?  Use made it what it is and
>> people are very unlikely to make major changes that break lots of already
>> written software.  That is what caused all the two year dates more than a
>> decade after punched-cards went away and keeps time_t too small and even
>> keeps IPV6 out of the mainstream (I first used IPV6 more than 10 years
>> ago and we actually had (have?) an address allocation).
> 
> I don't know the .TTYOUT library function that you used. Never seen it. 
> Looks very much like RT-11 to me. :-)

..TTYOUT isn't the offending operation.  That is a simple "output one
char to the terminal" and was used to do the CR-LF pairs. 
It is the .ASCIZ which creates a null-terminated string and the .PRINT
on a PDP-11 or .PUTSTR on VMS that display them that demonstrates that
it wasn't just a Unix concept.  It predated Unix (at least on the PDP-11)
and was used by other, supposedly safer, OSes like VMS after it was
adopted by Unix.  I am merely saying it was common practice in the
industry at the time that Unix adopted it and people need to stop
lating the blame at the feet of Unix (assuming there really is any blame
to lay.  I still say it is the programmers fault and not the language
or any particular facet of said language.)

> 
> Anyway, I bet that library is just as unsafe as any Unix code, yes. But 
> that library is rather in the face of VMS conventions, where you 
> actually have a string type defined, 

And, ther eis nothing to stop Unix or C programmers from doing the same
except for the fact that up to now, no one really seems to have cared.
Like I pointed out above.  The potential problems were known in the 70's
when both Unix and C were in their infancy.  Fixing them was rejected
by all the various parts of the community.

>                                      which should even be used if you 
> write in MACRO-32. And all system calls, as well as standard libraries 
> will expect that string type, and not your nul-terminated string, so you 
> will have to do the conversion every time you plan to use any standard 
> system functions or libraries.

Safe-C provided a string safe (supposedly) compatable library.  Never
having actually had the chance to use it, I don't know how well it worked.
Maybe what I need to do is try to track down the owner of the company and
see if any of it still exists.  I could always run some tests on Ultrix-11.
:-)

> And the standard string type in VMS have a pointer and a length field. 
> You should know this... :-)

I do know this.  And, there is no reason why Unix can't have the same other
than apathy.  Of course, the question still is which is actually the VMS
standard?  .ASCIZ did come first.

> 
>>> Despite .ASCIZ - find it used as part of the calling standard.
>>
>> The fact that it is still included as a part of the language tells me
>> is is, in fact, a "part of the calling standar" on VMS.  Just because
>> people choose to use something better doesn't remove it from the
>> standard.
> 
> No. The calling standard in VMS is well defined (in contradiction to 
> Unix systems, where it is not). Strings is their own type. Just because 
> you decide to use nul-terminated strings, and the fact that there are 
> libraries that accept these, don't make them the calling standard.

Only if one assumes "calling standards" are volatile.  In which case, the
Unix one could be changed.  Except that no one programming for Unix has
any interest in changing them.

> 
> I suspect that in the name of Unix/Posix compatibility, nul terminated 
> string have probably spread in the VMS world since I last played VMS 
> though, so maybe they are officially a part of the calling standard in 
> VMS nowadays. But I very much doubt they show up anywhere in the older 
> core of VMS.
> 
>>>                                                                   It was
>>> stupidity on a grand scale.
>>
>> Hindsight is always 20/20.  It served a purpose and worked very well for
>> a long time.  It is still usable and perfectly safe depending on the
>> programmer.   And, in some environments it is perfectly safe regardless.
>> I would foresee no problem using .ASCIZ cwunder RT-11, RSTS, RSX or under
>> VMS.  Of course, ~i have no problem with using it under Unix, but that's
>> me.  :-)
> 
> Using nul-terminated string is a potential disaster under any OS. No 
> system call under RSX accepts them, and very few library functions will 
> have anything to do with them.

So then, the RT-11 Programmed Requests supported under RSX are not
considered "system calls"?  Cause according to the RSX-11 manual
the RT-11 .PRINT Programmed Request is available on RSX.

> In RSX though, there is no system wide calling conventions. You have the 
> FORTRAN calling convention, using R5 as an argument pointer. You have 
> BASIC+2, which use another convention, you have PDP-11 C, which use yet 
> another. And they all represent strings in their own ways...
> Which also means that you need to convert strings whenever you mix 
> things, or when you do system calls.

No argument here.  All of this is really silly actually.  My point is
and always has been that neither C nor Unix invented null-terminated
strings.  When their use became ingrained in both the language and the
OS they were already common practice in the industry and their dangers
were already well known.

> 
> I try to stay away from nul-terminated string for most things, and only 
> use them when I'm lazy, and for very restricted and specific things.

And I have no problem using them because I have never made assumptions about
them and verify all uses.  Again, as I have said repeatedly.  It is the
programmers responsibility to make his program right and safe.  Blaming the
kanguage or the OS when the program screws up is ridiculous.  If you can't
handle programming with C or under Unix, just don;t do it!!!

And, as even more fuel for the fire, let's go back to something in more
current times.  Linux.  Started in 1991.  So then, why was it written in
C if C was such a bad language?  It was (supposedly) written from scratch
and not just a copy of an existing version of Unix.  Why did he choose
to write it in such an unsafe language?  Rhetorical question.  I think
everyone knows why.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8pppj$vsv$1@iltempo.update.uu.se>,
	Johnny Billquist <bqt@softjar.se> writes:
> On 2010-10-05 03:32, Michael Kraemer wrote:
>> In article<8guoccF6t8U1@mid.individual.net>, billg999@cs.uofs.edu (Bill
>> Gunshannon) writes:
>>>
>>> When are we going to put this myth to sleep.  Unix didn't invent null
>>> terminated strings.  DEC was perfectly happy with them as well.  All
>>> of the non-Unix PDP-11 OSes had them
>>
>> exactly.
>> The first time I encountered them (in a former life)
>> was on a PDP-11 under RSX-11-something.
>> It seemed to be the standard way to store varying length strings, IIRC.
>> Quite efficient IMHO for an architecture which maxes out at 16 bits,
>> where a 2-byte length prefix might be considered as a waste of resources
>> compared to a 1-byte trailing null byte.
> 
> Huh?
> I've been using RSX for over 20 years, and am still actively using it. 
> Nul-terminated strings are hardly the norm anywhere in the system.
> 
> Yes, you can (as always) create nul-terminated strings, and yes, 
> MACRO-11 even have an operation for it.
> However, no system call take that as an argument, and I can think of 
> about 1 library function that plays around with nul-terminated strings 
> ($EDMSG).
> 
> What some people might have done, when writing in assembler, to maximize 
> the memory usage is another thing, though. But don't point at RSX for that.

You can point at RSX just as well as Unix and C.  None of them imvented it.
It was common practice at the time.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <i8pqes$bk$1@Iltempo.Update.UU.SE>, Johnny Billquist <bqt@softjar.se>
writes:
> 
> I have not tried this, but as far as I read the man-page, I would 
> suspect that calling malloc(0) should return a pointer to a memory 
> address, but one where you don't have any bytes to play with. Like a 
> void pointer.
 
> It's like a malloc(0) should always succeed, since you can always 
> allocate 0 bytes of memory, those you will never run out of.
> 
> It would be weird to get a ENOMEM when you did a malloc(0).
> 

Sorry, these suggestions don't make sense for me.
malloc(0) is nonsense and almost always the consequence of a coding error.
Hence it should not be honoured by the RTL,
i.e. by returning a NULL pointer. 
What other pointer value refers to "bytes you don't have to play with"?
Fortunately this function is mostly implemented reasonably,
i.e. AIX, HP-UX and possibly others return NULL and errno=EINVAL
in the above case.

Johnny Billquist wrote:

> The problem with Unix is that system calls takes nul-terminated strings 
> as parameters.


strcpy and strcat are definitely "offending" routines. but there is
sntcvpy and strncat which are not "offending" because you do specify the
maximum size of the receiving buffer.

sprintf and scanf are also offending as they can easily overflow the
receiving buffer if the format specifier does not impose a limit.



BUT, most other "system" routines require a length argument be included
to limit the amount of data that can be written by the system service to
the receiving buffer.

On Sat, 09 Oct 2010 15:54:18 -0400, JF Mezei wrote:

> Johnny Billquist wrote:
> 
>> The problem with Unix is that system calls takes nul-terminated strings
>> as parameters.

True.

> strcpy and strcat are definitely "offending" routines. but there is
> sntcvpy and strncat which are not "offending" because you do specify the
> maximum size of the receiving buffer.
> 
> sprintf and scanf are also offending as they can easily overflow the
> receiving buffer if the format specifier does not impose a limit.

But these are not 'system' routines; they run at normal privilege level 
and are essentially part of the program.

If you had said 'open' or 'create', then yes...



-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

In article <i8pocb$ved$1@Iltempo.Update.UU.SE>, Johnny Billquist
<bqt@softjar.se> writes:
> 
> The problem with Unix is that system calls takes nul-terminated strings 
> as parameters. 

No it's not.
The vast majority of Unix/C programmers has no problems with it.
Just learn how to use it properly.

> And yes, most library calls do, too. It's a philosophical 
> concept. The fact that you can create nul-terminated strings in your 
> program is not a problem. The fact that you *have* to use nul-terminated 
> strings are...
> 

In article <4cb0c86c$0$18578$c3e8da3$38634283@news.astraweb.com>, JF Mezei
<jfmezei.spamnot@vaxination.ca> writes:
> Johnny Billquist wrote:
> 
> > The problem with Unix is that system calls takes nul-terminated strings 
> > as parameters.
> 
> 
> strcpy and strcat are definitely "offending" routines.

Not really.
Since one knows the lengths of the operands before,
one uses calloc() to create a buffer which always
has the right size. 

> but there is
> sntcvpy and strncat which are not "offending" because you do specify the
> maximum size of the receiving buffer.

But it doesn't provide the real solution to all
so-called buffer overflows: simply calloc() as much storage
as you need before writing to it. It is as simple as that,
no new/other languages needed here.
 
> sprintf and scanf are also offending as they can easily overflow the
> receiving buffer if the format specifier does not impose a limit.

sprintf() is indeed nasty because one has no easy means
to determine the total length of the target buffer before.
A construct like
nn = sprintf( NULL, "format-specs", ... );
str = calloc( nn+1, sizeof(str *) );
nn = sprintf( str , "format-specs", ... );
would be desirable, safe and still within the simple C logic. 
One may of course use dummy output to /dev/null
to get "nn", but that's a bit clumsy. 

> BUT, most other "system" routines require a length argument be included
> to limit the amount of data that can be written by the system service to
> the receiving buffer.

Again, really useful only if one alloc's the buffer before.

Michael Kraemer <m.kraemer@gsi.de> wrote:
(snip on malloc(0))
 
> Sorry, these suggestions don't make sense for me.
> malloc(0) is nonsense and almost always the consequence of a coding error.
> Hence it should not be honoured by the RTL,
> i.e. by returning a NULL pointer. 
> What other pointer value refers to "bytes you don't have to play with"?
> Fortunately this function is mostly implemented reasonably,
> i.e. AIX, HP-UX and possibly others return NULL and errno=EINVAL
> in the above case.

I could disagree, but that won't change the systems that are
out there.

Newer versions of Fortran allow for arrays (at least dynamically
allocated ones) of dimension zero.  It is a nice generalization
and saves some tests for special cases that are otherwise 
not needed.

Otherwise, most languages do not allow a dimension of zero,
and specifically C89 does not.  (Maybe changed in C99, I 
haven't looked.)

-- glen

On 2010-10-08 14:40, Simon Clubley wrote:
> On 2010-10-07, Rich Alderson<news@alderson.users.panix.com>  wrote:
>> Simon Clubley<clubley@remove_me.eisner.decus.org-Earth.UFP>  writes:
>>
>>> On 2010-10-06, Rich Alderson<news@alderson.users.panix.com>  wrote:
>>
>>>> News to the folks at XX2247, successors of Mentec.
>>
>>> What exactly do you mean by Mentec successor ?
>>
>>> Is this a successor in the legal sense of the word or a unrelated
>>> organisation ? Are they resuming Mentec's PDP-11 OS operations ?
>>
>> Yes, a legal successor.
>>
>> I thought they would have said something for themselves by now, at least on
>> their site.  I shouldn't have said anything, I guess.
>>
>
> Thanks for the feedback. Clearly there must still be a market for the PDP-11
> environment so it will be interesting to see what services they think are
> still viable to offer.

Well, I do know of several commercial installations of RSX, which is 
running as current software as there are...

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

On 2010-10-09 01:35, glen herrmannsfeldt wrote:
> Rich Alderson<news@alderson.users.panix.com>  wrote:
> (snip, I wrote)
>
>>> There seems to be a xx2247.org, but isn't org supposed
>>> to be for non-profit organizations?
>
>> Not for a very very long time now.
>
> If you mean that there is no enforcement for the distinction,
> then I completely agree.  Now, I suppose some might use the .org
> domain when the matching .com was already taken, but that doesn't
> seem to be true here.
>
> Since many people assume .com, without always reading the URL,
> I don't see why anyone with a profit-making company would register
> just the .org.

Good point, but as Rich pointed out, there is no guarantees about this. 
I guess we'll have to wait and see.

>> There also used to a division among IP addresses among class A (1.n.n.n to
>> 127.n.n.n), class B (128.n.n.n to 191.n.n.n) and class C (192.n.n.n to
>> 254.n.n.n).  You do know that that changed as well, right?
>
> Yes, I know about that one.
>
> I thought class C only went up to 223, do leave room for future classes.
> Possibly that was added later, though.

No, you are correct. There is a class D as well, which is 224 to 239. 
This is perhaps more known as the IP multicast addresses.
Class E is what is reserved for future growth (240-255).

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

"Bill Gunshannon" <billg999@cs.uofs.edu> wrote in message 
news:8hb9i2FgnoU1@mid.individual.net...
> In article <i8pms6$v0t$1@iltempo.update.uu.se>,
> Johnny Billquist <bqt@softjar.se> writes:
>> On 2010-10-08 23:23, JF Mezei wrote:
>>> Bill Gunshannon wrote:
>>>
>>>> Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
>>>> contains a lot of GNU) running on top of it.  The only Apple 
>>>> contribution
>>>> is their wierd window manager.
>>>
>>> Apple did not just "use" Mach. It fixed it. There were good reason why
>>> it had not been widely adopted in the past. Apple fixed those reasons.
>>> (I rememer reading an article about this, but the technical aspect elude
>>> me at the moment).
>>
>> Well, technically, OS-X isn't much Mach at all. It's mostly NextSTEP
>> with FreeBSD/NetBSD thrown in. NextSTEP in turn was based on Mach,
>> people say, but that is in a very limited way, as NextSTEP was also BSD
>> based, with just a few cherry-picked Mach:isms thrown in.
>> A serious Mach lover would not recognize himself in OS-X.
>>
>> I know that everyone says that OS-X is Mach based, but I'd like anyone
>> to show me where. The Mach kernel was based around a few simple
>> concepts. It's a microkernel, with an extensive IPC having features and
>> possibilities not existing in more classical Unixes. Neither of these
>> two features have been adopted by anything else.
>> The third part is the VM system, which you might argue all modern BSD
>> based systems use, which was adopted from Mach.
>>
>> So, no. OS-X is definitely not the Mach kernel with the BSD Userland.
>
> OK, I'll concede that although it really appears to depend on where you
> read.
>
>> It's mostly BSD kernel with BSD userland.
>
> Which would make it even more just another distro of Unix.
>
>>                                            If you can find anything of
>> Mach in there, please let me know.
>>
>> There is quite a lot of information in Wikipedia about Mach, OS-X,
>> NextSTEP and XNU, for anyone to read if they are interested.
>
> I am  not inclined to tear it appart to see what's at the guts.  It's
> a religion anyway and I am not likely to conbince anyone anyway.
>
>>
>> That said, someone else claimed that Linux was bigger than OS-X. That
>> don't seem to be true. OS-X as a Unix system overshadows all other Unix
>> variants combined, when it comes to deployed systems (yes, add all
>> Linux, Solaris, xBSD, OSF/1, Ultrix, whatever together, and it's fewer
>> systems than the OS-X ones...)
>
> Well, I assume you are getting these numbers by counting iPhones and
> other devices with embeded OS-X.  If that's the case yopu really need
> to count all the systems with embedded versions of the other OSes as
> well.
>
> How many Netgear switches have been sold world-wide?
> How many Linksys Access Points?
> How many Linksys/Cisco NSS boxes?
>
> All of them have Linux under the hood.  And there are hundreds more
> each having sold thousands of units.
>
> The original argument was about desktop usage as opposed to server
> usage and wasn't really about embeded systems at all.
>
> bill
>
> -- 
> Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
> billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
> University of Scranton   |
> Scranton, Pennsylvania   |         #include <std.disclaimer.h>

"How many Netgear switches have been sold world-wide?

How many Linksys Access Points?

How many Linksys/Cisco NSS boxes?

All of them have Linux under the hood. And there are hundreds more

each having sold thousands of units."

Sorry Bill but that's plain wrong.

Wrt switches: An unmanaged switch, regardless of vendor, is likely to not 
even have an OS. Unmanaged switches are the vast majority of switches sold. 
None of them have Linux under the hood; most of them are low-cost 
single-chip things which don't even have a processor int the conventional 
sense, let alone an OS.

Wrt access points: Linksys used to make a famous Access Point with Linux 
under the hood (the WRT54G), and it was popular target for people who wanted 
a hardware platform for a Linux-based whatever. Others have done Linux in 
this kind of kit too, in their DSL routers and so on.

Linux is bigger than a dedicated embedded OS, needs more ROM and RAM. 
Sometimes there are worthwhile cost savings to be had by moving to a more 
compact dedicated OS. Which is why Linksys WRT54 family now ship with (now 
Intel-owned) VxWorks, a commercial OS which is far from cheap to licence, 
but which has allowed Linksys to halve the ROM and RAM in the relevant APs. 
Presumably the per-unit cost saving from that outweighs the per-unit extra 
cost of VxWorks - you can get VxWorks as a one-time per-project licence 
which you pay (lots) for up front and there are then no per-box costs.

And going back a little further...

"have we all forgotten things line "Safe C" which was created back in the 
70's and as a commercial product, didn't fly."

What's your point? That you've not followed Safe C principles and usage 
since the mid 1990s?

That particular compiler and runtime may no longer exist, but the principles 
on which it was based continue to be the basis for things like the MISRA 
(Motor Industry Software Reliability Association) C and associated rules, 
documents and tools. Sensible motor manufacturers (which isn't all of them) 
and/or their suppliers follow these, as do sensible system builders outside 
the motor industry. You can get the benefit of a compiler that's in 
widespread use so other folks can do lots of testing of the compiler/chip 
pair, and the benefit of a set of design and coding guidelines and tools 
intended to lead to better code. In general it's preferable to use a widely 
used compiler/chip pair rather than using an obscure compiler,runtime and 
chip used by a tiny proportion of the C market (hello MIL1750).

There's also MISRA for C++. A supplier with *only* a C compiler and tools 
and no support for C++ leaves themselves rather behind the market, even in 
the safety-critical market. It's not a small upgrade, so it's really no 
great surprise that Safe C disappeared.

Summary: It is wrong to assume that, simply because "Safe C" went away, 
there is insufficient interest in the topic to justify continued commercial 
interest in the topic.

In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>, "John Wallace" <not@any.price> writes:
>
>Wrt access points: Linksys used to make a famous Access Point with Linux 
>under the hood (the WRT54G), and it was popular target for people who wanted 
>a hardware platform for a Linux-based whatever. Others have done Linux in 
>this kind of kit too, in their DSL routers and so on.
>

Linksys made an "infamous" Access Point too, the WAP54G.  Once, or twice or
even thrice a day, it would need to be rebooted as it froze up.  Reboot, of
course, meaning a power cycle because its web based controls were inacess-
ible when the units froze.  I'm still suffering one.  I need to find a new
"access point" (not wireless router) one of these days.

-- 
VAXman- A Bored Certified VMS Kernel Mode Hacker    VAXman(at)TMESIS(dot)ORG

All your spirit rack abuses, come to haunt you back by day.
All your Byzantine excuses, given time, given you away.

On Sun, 10 Oct 2010 13:13:35 +0000, VAXman- wrote:

> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>, "John
> Wallace" <not@any.price> writes:
>>
>>Wrt access points: Linksys used to make a famous Access Point with Linux
>>under the hood (the WRT54G), and it was popular target for people who
>>wanted a hardware platform for a Linux-based whatever. Others have done
>>Linux in this kind of kit too, in their DSL routers and so on.
>>
>>
> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or twice
> or even thrice a day, it would need to be rebooted as it froze up. 
> Reboot, of course, meaning a power cycle because its web based controls
> were inacess- ible when the units froze.  I'm still suffering one.  I
> need to find a new "access point" (not wireless router) one of these
> days.

Strange. I have a WAP54G here, had it for a long time (probably since 
they first came out).

It's on 24/7 and has never locked up. It's only been rebooted to change 
its IP address and other reconfiguration.



-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>,
	"John Wallace" <not@any.price> writes:
> 
> "Bill Gunshannon" <billg999@cs.uofs.edu> wrote in message 
> news:8hb9i2FgnoU1@mid.individual.net...
>> In article <i8pms6$v0t$1@iltempo.update.uu.se>,
>> Johnny Billquist <bqt@softjar.se> writes:
>>> On 2010-10-08 23:23, JF Mezei wrote:
>>>> Bill Gunshannon wrote:
>>>>
>>>>> Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
>>>>> contains a lot of GNU) running on top of it.  The only Apple 
>>>>> contribution
>>>>> is their wierd window manager.
>>>>
>>>> Apple did not just "use" Mach. It fixed it. There were good reason why
>>>> it had not been widely adopted in the past. Apple fixed those reasons.
>>>> (I rememer reading an article about this, but the technical aspect elude
>>>> me at the moment).
>>>
>>> Well, technically, OS-X isn't much Mach at all. It's mostly NextSTEP
>>> with FreeBSD/NetBSD thrown in. NextSTEP in turn was based on Mach,
>>> people say, but that is in a very limited way, as NextSTEP was also BSD
>>> based, with just a few cherry-picked Mach:isms thrown in.
>>> A serious Mach lover would not recognize himself in OS-X.
>>>
>>> I know that everyone says that OS-X is Mach based, but I'd like anyone
>>> to show me where. The Mach kernel was based around a few simple
>>> concepts. It's a microkernel, with an extensive IPC having features and
>>> possibilities not existing in more classical Unixes. Neither of these
>>> two features have been adopted by anything else.
>>> The third part is the VM system, which you might argue all modern BSD
>>> based systems use, which was adopted from Mach.
>>>
>>> So, no. OS-X is definitely not the Mach kernel with the BSD Userland.
>>
>> OK, I'll concede that although it really appears to depend on where you
>> read.
>>
>>> It's mostly BSD kernel with BSD userland.
>>
>> Which would make it even more just another distro of Unix.
>>
>>>                                            If you can find anything of
>>> Mach in there, please let me know.
>>>
>>> There is quite a lot of information in Wikipedia about Mach, OS-X,
>>> NextSTEP and XNU, for anyone to read if they are interested.
>>
>> I am  not inclined to tear it appart to see what's at the guts.  It's
>> a religion anyway and I am not likely to conbince anyone anyway.
>>
>>>
>>> That said, someone else claimed that Linux was bigger than OS-X. That
>>> don't seem to be true. OS-X as a Unix system overshadows all other Unix
>>> variants combined, when it comes to deployed systems (yes, add all
>>> Linux, Solaris, xBSD, OSF/1, Ultrix, whatever together, and it's fewer
>>> systems than the OS-X ones...)
>>
>> Well, I assume you are getting these numbers by counting iPhones and
>> other devices with embeded OS-X.  If that's the case yopu really need
>> to count all the systems with embedded versions of the other OSes as
>> well.
>>
>> How many Netgear switches have been sold world-wide?
>> How many Linksys Access Points?
>> How many Linksys/Cisco NSS boxes?
>>
>> All of them have Linux under the hood.  And there are hundreds more
>> each having sold thousands of units.
>>
>> The original argument was about desktop usage as opposed to server
>> usage and wasn't really about embeded systems at all.
>>
> "How many Netgear switches have been sold world-wide?
> 
> How many Linksys Access Points?
> 
> How many Linksys/Cisco NSS boxes?
> 
> All of them have Linux under the hood. And there are hundreds more
> 
> each having sold thousands of units."
> 
> Sorry Bill but that's plain wrong.
> 
> Wrt switches: An unmanaged switch, regardless of vendor, is likely to not 
> even have an OS. 

I should have been clearer.  I meant managed switches.

>                   Unmanaged switches are the vast majority of switches sold. 
> None of them have Linux under the hood; most of them are low-cost 
> single-chip things which don't even have a processor int the conventional 
> sense, let alone an OS.

In homes, probably, but there are a lot of places using managed Netgear
switches as opposed to Cisco.  I know because we have been using them
for years.

> 
> Wrt access points: Linksys used to make a famous Access Point with Linux 
> under the hood (the WRT54G), and it was popular target for people who wanted 
> a hardware platform for a Linux-based whatever. Others have done Linux in 
> this kind of kit too, in their DSL routers and so on.
> 
> Linux is bigger than a dedicated embedded OS, needs more ROM and RAM. 
> Sometimes there are worthwhile cost savings to be had by moving to a more 
> compact dedicated OS. 

Size is irrelevant to the discusion at hand.  I still hold that if you
count all the things that run Linux they outnumber all the things running
OS-X. And you can't even count all the instances of Linux becuae you
don't pay for them so you can't count and you don't register free installs.
I have at least a dozen machines at home runnng many variants of Linux
and not one of them shows up in any count.

>                       Which is why Linksys WRT54 family now ship with (now 
> Intel-owned) VxWorks, a commercial OS which is far from cheap to licence, 
> but which has allowed Linksys to halve the ROM and RAM in the relevant APs. 
> Presumably the per-unit cost saving from that outweighs the per-unit extra 
> cost of VxWorks - you can get VxWorks as a one-time per-project licence 
> which you pay (lots) for up front and there are then no per-box costs.

I expect the real reason for LinkSys dumping Linux was to stop the third
party firmware world from creating hacked versions of their boxes.

> 
> And going back a little further...
> 
> "have we all forgotten things line "Safe C" which was created back in the 
> 70's and as a commercial product, didn't fly."
> 
> What's your point? That you've not followed Safe C principles and usage 
> since the mid 1990s?

I follow them all the time which is why I don't have a problem with C
or Unix. (Remember me, I'm the one who says the problem is the programmer,
not the language or the OS!)  My point was that the industry had an 
alternative over 30 years ago.  It resoundingly rejected it.  Of course,
back then programming was done by programmers and not by 13 year olds
claiming to be IT Professionals.

> 
> That particular compiler and runtime may no longer exist, but the principles 
> on which it was based continue to be the basis for things like the MISRA 
> (Motor Industry Software Reliability Association) C and associated rules, 
> documents and tools. Sensible motor manufacturers (which isn't all of them) 
> and/or their suppliers follow these, as do sensible system builders outside 
> the motor industry. You can get the benefit of a compiler that's in 
> widespread use so other folks can do lots of testing of the compiler/chip 
> pair, and the benefit of a set of design and coding guidelines and tools 
> intended to lead to better code. In general it's preferable to use a widely 
> used compiler/chip pair rather than using an obscure compiler,runtime and 
> chip used by a tiny proportion of the C market (hello MIL1750).
> 
> There's also MISRA for C++. A supplier with *only* a C compiler and tools 
> and no support for C++ leaves themselves rather behind the market, even in 
> the safety-critical market. It's not a small upgrade, so it's really no 
> great surprise that Safe C disappeared.
> 
> Summary: It is wrong to assume that, simply because "Safe C" went away, 
> there is insufficient interest in the topic to justify continued commercial 
> interest in the topic.
 
That one facet of the IT industry has decided to be concerned doesn't
discount my argument.  "insufficient interest" merely says that the
majority do not care.  If they did the C library as used in all those
free Unix systems today would have been rewritten in a safe way. It
is not like they have not made changes along the way that were not
backwards compatable.  But, they have no interest in doing so.

By the way, what would it cost to actually get a copy of any of their
documents?  If MISRA is like other "standards" bodies (IEEE for example)
their cost for providing any of this information is prohibitive.  One
would think that widespread distribution of standards would be seen as
a good thing and a primary purpose of a standards body, but reality is
rather otherwise!!

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <00AA4BE3.E1850FC7@sendspamhere.org>,
	VAXman-  @SendSpamHere.ORG writes:
> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>, "John Wallace" <not@any.price> writes:
>>
>>Wrt access points: Linksys used to make a famous Access Point with Linux 
>>under the hood (the WRT54G), and it was popular target for people who wanted 
>>a hardware platform for a Linux-based whatever. Others have done Linux in 
>>this kind of kit too, in their DSL routers and so on.
>>
> 
> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or twice or
> even thrice a day, it would need to be rebooted as it froze up.  Reboot, of
> course, meaning a power cycle because its web based controls were inacess-
> ible when the units froze.  I'm still suffering one.  I need to find a new
> "access point" (not wireless router) one of these days.

You seem to have problems with everythng you use.  I, on the other hand,
have a network of WAP54G's running through the building from our department
are to the classroom area so that faculty can just pickup their laptops
and walk between their offices and the classrooms while maintaining
connectivity.  System has been in place since the initial release of the
WAP54G (I really don't remember when we first installed them but it was
at least 4 years ago). The only time they get rebooted is when the power
in the building fails and since the introduction of POE adapters they have
run off one power connection with a UPS so that power failures are non-
existant.  They never lockup and I never reboot them.  They have run like
this thru 2 6ix-month periods when I was away witht he Army without a
single failure.  Are you sure you are not really Joe Btfsplk?

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

On Sun, 10 Oct 2010 16:17:45 +0000, Bill Gunshannon wrote:

> By the way, what would it cost to actually get a copy of any of their
> documents?  If MISRA is like other "standards" bodies (IEEE for example)
> their cost for providing any of this information is prohibitive.  One
> would think that widespread distribution of standards would be seen as a
> good thing and a primary purpose of a standards body, but reality is
> rather otherwise!!

I agree entirely, not just with this standard but with a lot of them. 
However, my solution was:

 http://tinyurl.com/c99standard




-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

On 2010-10-09 15:38, Bill Gunshannon wrote:
> In article<i8pms6$v0t$1@iltempo.update.uu.se>,
> 	Johnny Billquist<bqt@softjar.se>  writes:
>> On 2010-10-08 23:23, JF Mezei wrote:
>>> Bill Gunshannon wrote:
>>>
>>>> Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
>>>> contains a lot of GNU) running on top of it.  The only Apple contribution
>>>> is their wierd window manager.
>>>
>>> Apple did not just "use" Mach. It fixed it. There were good reason why
>>> it had not been widely adopted in the past. Apple fixed those reasons.
>>> (I rememer reading an article about this, but the technical aspect elude
>>> me at the moment).
>>
>> Well, technically, OS-X isn't much Mach at all. It's mostly NextSTEP
>> with FreeBSD/NetBSD thrown in. NextSTEP in turn was based on Mach,
>> people say, but that is in a very limited way, as NextSTEP was also BSD
>> based, with just a few cherry-picked Mach:isms thrown in.
>> A serious Mach lover would not recognize himself in OS-X.
>>
>> I know that everyone says that OS-X is Mach based, but I'd like anyone
>> to show me where. The Mach kernel was based around a few simple
>> concepts. It's a microkernel, with an extensive IPC having features and
>> possibilities not existing in more classical Unixes. Neither of these
>> two features have been adopted by anything else.
>> The third part is the VM system, which you might argue all modern BSD
>> based systems use, which was adopted from Mach.
>>
>> So, no. OS-X is definitely not the Mach kernel with the BSD Userland.
>
> OK, I'll concede that although it really appears to depend on where you
> read.

:-)
If you ever take a step back when using a MAC, you'll easily see that 
it's not Mach. It is not a micro kernel, but follows the traditional 
Unix monolithic kernel philosophy. It also do IPC the same way as any 
other Unix, and don't use message passing for everything (including 
system calls) like Mach do (did).

Mach was for a while very hip and cool to associate yourself with, even 
though noone really did adopt most of it.

Nowadays, Mach is pretty dead.

>> It's mostly BSD kernel with BSD userland.
>
> Which would make it even more just another distro of Unix.

Indeed! And it is... The only Apple-specific thing is the GUI, and the 
semi-"get-rid-of text-readable-configuration-files". Apple in all their 
wisdom implemented a hellish binary configuration data structure 
instead, so that you need to use their tools to change a lot of things, 
instead of just poking in some nice text files.

>>                                             If you can find anything of
>> Mach in there, please let me know.
>>
>> There is quite a lot of information in Wikipedia about Mach, OS-X,
>> NextSTEP and XNU, for anyone to read if they are interested.
>
> I am  not inclined to tear it appart to see what's at the guts.  It's
> a religion anyway and I am not likely to conbince anyone anyway.

:-)
I don't think we need either. I believe your original claim was that OS 
X was just something Apple grabbed from somewhere else and repackaged. I 
think we can all agree on this, just as we can agree that it's just 
another Unix...

>> That said, someone else claimed that Linux was bigger than OS-X. That
>> don't seem to be true. OS-X as a Unix system overshadows all other Unix
>> variants combined, when it comes to deployed systems (yes, add all
>> Linux, Solaris, xBSD, OSF/1, Ultrix, whatever together, and it's fewer
>> systems than the OS-X ones...)
>
> Well, I assume you are getting these numbers by counting iPhones and
> other devices with embeded OS-X.  If that's the case yopu really need
> to count all the systems with embedded versions of the other OSes as
> well.

No, I was just talking about normal computers. The most obvious number 
being web clients, which is very popular to collect statistics on. And 
those outnumber servers by such a wide margin that servers don't show on 
the radar anyway.

If we start counting embedded systems, I think the numbers might look 
different, but I'm not sure they are different enough to be interesting.

I know of a few systems where Linux is embedded, but I have not seen a 
single one where production are counted in the millions.

But I'm sure you can find some, if you search. However, most embedded 
systems run more dedicated OSes.

> How many Netgear switches have been sold world-wide?
> How many Linksys Access Points?
> How many Linksys/Cisco NSS boxes?
>
> All of them have Linux under the hood.  And there are hundreds more
> each having sold thousands of units.
>
> The original argument was about desktop usage as opposed to server
> usage and wasn't really about embeded systems at all.

Yeah. And desktop Linux is very small compared to OS X. But on the other 
hand, in my eyes, they are all Unix, and are all in a small minority to 
Windows. :-)

See http://en.wikipedia.org/wiki/Usage_share_of_operating_systems for 
some numbers. But to make it short:

Windows XP (48.32%)
Windows 7 (19.81%)
Windows Vista (18.43%)
Mac OS X (6.42%)
iOS (iPhone) (1.40%)
Linux (1.34%)

Meaning that iPhone alone is more common than Linux, and OS X is more 
than 4 times as common as Linux.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

In article <i8pr8r$gs$1@iltempo.update.uu.se>,
	Johnny Billquist <bqt@softjar.se> writes:
> On 2010-10-05 19:56, Bill Gunshannon wrote:
>> In article<i8dq16$d9r$1@usenet01.boi.hp.com>,
>> 	"FredK"<fred.nospam@dec.com>  writes:
>>>
>>> Thanks.  Essentially because certain routines didn't include an error
>>> return, they created the errno hack.  Quite a hack... a global symbol.
>>
>> Matter of opinion.  It has served Unix programmers well for a lot longer
>> than VMS has even been around.
> 
> The same solution exists in most OSes. In VMS, it is instead returned in 
> R0 from system calls, which is better since it can easily be made thread 
> safe. But it is still just a value set somewhere to tell whether the 
> call succeeded or not.
> 
> In RSX, the value is stored in $DSW, which serves exactly the same 
> purpose as errno in Unix.
> 
>>> I can defend things like C not having a general string type if C is viewed
>>> as a assembler replacement as opposed to a high-level language (although a
>>> counted string would make more sense in general than null terminated).
>>
>> If you want a counted string, create it.  That's one of the beauties of
>> C.  The basic types as might be defined at the hardware level already
>> exist and you are free to create any other user defined type you feel
>> necessary.  Maker a string what you want a string to be.  I have worked
>> with counted strings (ala UCSD Pascal) I saw no real advantage to them.
>> And one very big disadvantage.  They come with a very finite fixed maximum
>> length.  If nothing else, that either keeps them very short or makes them
>> very non-portable.
> 
> One of the problems with this in Unix is that you need to convert it 
> back to a nul-terminated string for system calls and library calls. 
> Which is why it is a problem in Unix. Unix does not appreciate counted 
> strings.

But, being as Unix has been available in source for nigh on to 40 years,
this could have been fixed.  Why was it not?  Thus my statement that the
real problem was apathy.  No one in the Unix community was concerned
enough about it to fix the problem.

> 
> As for your experience with Pascal, I don't see the relevance. There is 
> no need to place the string length in the first byte of the string, and 
> limit string lengths to 255 bytes. That is just stupid, and it is not 
> the only alternative.

I onlt picked Pascal because UCSD-Pascal was problebly one of the earlier
and best know systems with counted strings.  It doesn't matter what the
size is, it is a fixed limitationi that kind of string and if we go back
to when C was coming into being, anything beyond a 255 byte string would
have required using more than one, very rare and valuable, byte to store
that length.  Null terminated always used one byte of overhead no matter
what the size of the string was.  Today, this is no longer a restriction.
So, why don't we fix it?  

> 
>>> The language decision led to OS API decisions.
>>
>> How?  The language does not define a "string" type.  That is actually
>> what is in the OS API.  And there have been numerous oportunities to
>> change it.  No one showed any interest. (Or have we all forgotten things
>> line "Safe C" which was created back in the 70's and as a commercial
>> product, didn't fly.)
> 
> You cannot change the system calls on a whim. Everything will break.

Various versions of Unix have always broken something.  How much code
being written toay will still compile and run on Ultrix (11 or 32)?
How about on Version-7?  Or, for that matter, how may of the ports
in the FreeBSD 8.0 ports tree will compile and run on FreeBSD 6.0?
Or 4.0?  There is nothing stopping the controlling developers from
drawing a line in the stand and saying this is where we fix things.
Nothing beyond lack of interest, that is.

> And that is the root of Unix problem. The API is not well though out 
> from a safety point of view, and we're stuck with it.

As has been said repeatedly, hindsight is always 20/20.  There was
nothing wrong with the API when Unix was created.  Because of the
state of society at the time the threats did not exist.  Because
there actually was a distinction between real programmers and pretenders
the threat of accidents was much less.  And because of the lack of
resources frugality was necessary in the basic design.  All of this
has changed today and the problem really should be fixed.  I am not
going to hold my breath waiting for that to happen.  But I will continue
to fight the arguments that somehow Unix was thrown together with no
design plan by incompetents.  I know better.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <8he7mnFr4lU2@mid.individual.net>, billg999@cs.uofs.edu (Bill Gunshannon) writes:
>In article <00AA4BE3.E1850FC7@sendspamhere.org>,
>	VAXman-  @SendSpamHere.ORG writes:
>> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>, "John Wallace" <not@any.price> writes:
>>>
>>>Wrt access points: Linksys used to make a famous Access Point with Linux 
>>>under the hood (the WRT54G), and it was popular target for people who wanted 
>>>a hardware platform for a Linux-based whatever. Others have done Linux in 
>>>this kind of kit too, in their DSL routers and so on.
>>>
>> 
>> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or twice or
>> even thrice a day, it would need to be rebooted as it froze up.  Reboot, of
>> course, meaning a power cycle because its web based controls were inacess-
>> ible when the units froze.  I'm still suffering one.  I need to find a new
>> "access point" (not wireless router) one of these days.
>
>You seem to have problems with everythng you use.  I, on the other hand,
>have a network of WAP54G's running through the building from our department
>are to the classroom area so that faculty can just pickup their laptops
>and walk between their offices and the classrooms while maintaining
>connectivity.  System has been in place since the initial release of the
>WAP54G (I really don't remember when we first installed them but it was
>at least 4 years ago). The only time they get rebooted is when the power
>in the building fails and since the introduction of POE adapters they have
>run off one power connection with a UPS so that power failures are non-
>existant.  They never lockup and I never reboot them.  They have run like
>this thru 2 6ix-month periods when I was away witht he Army without a
>single failure.  Are you sure you are not really Joe Btfsplk?

Here's the firmware rev.  

Firmware Version:  	v3.04, Dec 27, 2007

Perhaps, you have the version of the code that works?  I'd love to know.
The rest of the networking gear here is Cisco.  I'd just like a WAP that
is reliable.  Presently, it is so UNRELIABLE that it is plugged into one
of my APC MasterSwitch outlets so that it can SNMP power-cycled remotely.

FWIW, the firmware site for WAP54G is in .NZ and is NO LONGER accessible.

There's reportedly firmware update V3.1 and a V3.04.03 version.

-- 
VAXman- A Bored Certified VMS Kernel Mode Hacker    VAXman(at)TMESIS(dot)ORG

All your spirit rack abuses, come to haunt you back by day.
All your Byzantine excuses, given time, given you away.

On Sun, 10 Oct 2010 18:27:23 +0000, VAXman- wrote:

> In article <8he7mnFr4lU2@mid.individual.net>, billg999@cs.uofs.edu (Bill
> Gunshannon) writes:
>>In article <00AA4BE3.E1850FC7@sendspamhere.org>,
>>	VAXman-  @SendSpamHere.ORG writes:
>>> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>, "John
>>> Wallace" <not@any.price> writes:
>>>>
>>>>Wrt access points: Linksys used to make a famous Access Point with
>>>>Linux under the hood (the WRT54G), and it was popular target for
>>>>people who wanted a hardware platform for a Linux-based whatever.
>>>>Others have done Linux in this kind of kit too, in their DSL routers
>>>>and so on.
>>>>
>>>>
>>> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or
>>> twice or even thrice a day, it would need to be rebooted as it froze
>>> up.  Reboot, of course, meaning a power cycle because its web based
>>> controls were inacess- ible when the units froze.  I'm still suffering
>>> one.  I need to find a new "access point" (not wireless router) one of
>>> these days.
>>
>>You seem to have problems with everythng you use.  I, on the other hand,
>>have a network of WAP54G's running through the building from our
>>department are to the classroom area so that faculty can just pickup
>>their laptops and walk between their offices and the classrooms while
>>maintaining connectivity.  System has been in place since the initial
>>release of the WAP54G (I really don't remember when we first installed
>>them but it was at least 4 years ago). The only time they get rebooted
>>is when the power in the building fails and since the introduction of
>>POE adapters they have run off one power connection with a UPS so that
>>power failures are non- existant.  They never lockup and I never reboot
>>them.  They have run like this thru 2 6ix-month periods when I was away
>>witht he Army without a single failure.  Are you sure you are not really
>>Joe Btfsplk?
> 
> Here's the firmware rev.
> 
> Firmware Version:  	v3.04, Dec 27, 2007
> 
> Perhaps, you have the version of the code that works?  I'd love to know.
> The rest of the networking gear here is Cisco.  I'd just like a WAP that
> is reliable.  Presently, it is so UNRELIABLE that it is plugged into one
> of my APC MasterSwitch outlets so that it can SNMP power-cycled
> remotely.
> 
> FWIW, the firmware site for WAP54G is in .NZ and is NO LONGER
> accessible.
> 
> There's reportedly firmware update V3.1 and a V3.04.03 version.

Mine is "2005, Dec 28, ETSI, v3.05"

-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

On Sun, 10 Oct 2010 18:49:51 +0200, Johnny Billquist wrote:

> I know of a few systems where Linux is embedded, but I have not seen a
> single one where production are counted in the millions.

My ROM/FPGA/etc. programmer runs Linux!

-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

On Sun, 10 Oct 2010 18:27:23 +0000, VAXman- wrote:

> In article <8he7mnFr4lU2@mid.individual.net>, billg999@cs.uofs.edu (Bill
> Gunshannon) writes:
>>In article <00AA4BE3.E1850FC7@sendspamhere.org>,
>>	VAXman-  @SendSpamHere.ORG writes:
>>> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>, "John
>>> Wallace" <not@any.price> writes:
>>>>
>>>>Wrt access points: Linksys used to make a famous Access Point with
>>>>Linux under the hood (the WRT54G), and it was popular target for
>>>>people who wanted a hardware platform for a Linux-based whatever.
>>>>Others have done Linux in this kind of kit too, in their DSL routers
>>>>and so on.
>>>>
>>>>
>>> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or
>>> twice or even thrice a day, it would need to be rebooted as it froze
>>> up.  Reboot, of course, meaning a power cycle because its web based
>>> controls were inacess- ible when the units froze.  I'm still suffering
>>> one.  I need to find a new "access point" (not wireless router) one of
>>> these days.
>>
>>You seem to have problems with everythng you use.  I, on the other hand,
>>have a network of WAP54G's running through the building from our
>>department are to the classroom area so that faculty can just pickup
>>their laptops and walk between their offices and the classrooms while
>>maintaining connectivity.  System has been in place since the initial
>>release of the WAP54G (I really don't remember when we first installed
>>them but it was at least 4 years ago). The only time they get rebooted
>>is when the power in the building fails and since the introduction of
>>POE adapters they have run off one power connection with a UPS so that
>>power failures are non- existant.  They never lockup and I never reboot
>>them.  They have run like this thru 2 6ix-month periods when I was away
>>witht he Army without a single failure.  Are you sure you are not really
>>Joe Btfsplk?
> 
> Here's the firmware rev.
> 
> Firmware Version:  	v3.04, Dec 27, 2007
> 
> Perhaps, you have the version of the code that works?  I'd love to know.
> The rest of the networking gear here is Cisco.  I'd just like a WAP that
> is reliable.  Presently, it is so UNRELIABLE that it is plugged into one
> of my APC MasterSwitch outlets so that it can SNMP power-cycled
> remotely.
> 
> FWIW, the firmware site for WAP54G is in .NZ and is NO LONGER
> accessible.
> 
> There's reportedly firmware update V3.1 and a V3.04.03 version.

Sigh. We have to help you out again...this URL was found VIA linksys.com, 
and was obvious. Unless .com has moved to New Zealand...

 http://homesupport.cisco.com/en-us/wireless/lbc/WAP54G

-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

In article <8he8e1FvkjU13@mid.individual.net>,
	Bob Eager <rde42@spamcop.net> writes:
> On Sun, 10 Oct 2010 16:17:45 +0000, Bill Gunshannon wrote:
> 
>> By the way, what would it cost to actually get a copy of any of their
>> documents?  If MISRA is like other "standards" bodies (IEEE for example)
>> their cost for providing any of this information is prohibitive.  One
>> would think that widespread distribution of standards would be seen as a
>> good thing and a primary purpose of a standards body, but reality is
>> rather otherwise!!
> 
> I agree entirely, not just with this standard but with a lot of them. 
> However, my solution was:
> 
>  http://tinyurl.com/c99standard
 
Well, I didn't say all standards.  And, I asked about MISRA.  But let me
justify my argument by mentioning "OpenPROM".  Created by Sun.  Turned
over to IEEE and locked away with a pricetag few could afford.

Why do I know?  A couple of years ago when people were talking about how
hard it was to get boot code for a lot of PDP-11 devices I gave serious
thought to looking into an OpenPROM monitor for the PDP-11.  Until I saw
how much it would cost just to look at it.  And my employer is an IEEE
member!!!

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

On Sun, 10 Oct 2010 18:59:17 +0000, Bill Gunshannon wrote:

> In article <8he8e1FvkjU13@mid.individual.net>,
> 	Bob Eager <rde42@spamcop.net> writes:
>> On Sun, 10 Oct 2010 16:17:45 +0000, Bill Gunshannon wrote:
>> 
>>> By the way, what would it cost to actually get a copy of any of their
>>> documents?  If MISRA is like other "standards" bodies (IEEE for
>>> example) their cost for providing any of this information is
>>> prohibitive.  One would think that widespread distribution of
>>> standards would be seen as a good thing and a primary purpose of a
>>> standards body, but reality is rather otherwise!!
>> 
>> I agree entirely, not just with this standard but with a lot of them.
>> However, my solution was:
>> 
>>  http://tinyurl.com/c99standard
>  
> Well, I didn't say all standards.  And, I asked about MISRA.  But let me
> justify my argument by mentioning "OpenPROM".  Created by Sun.  Turned
> over to IEEE and locked away with a pricetag few could afford.
> 
> Why do I know?  A couple of years ago when people were talking about how
> hard it was to get boot code for a lot of PDP-11 devices I gave serious
> thought to looking into an OpenPROM monitor for the PDP-11.  Until I saw
> how much it would cost just to look at it.  And my employer is an IEEE
> member!!!

I know. It's dreadful, and no better here in Europe.

-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

In article <i8sqrg$cit$1@iltempo.update.uu.se>,
	Johnny Billquist <bqt@softjar.se> writes:
> On 2010-10-09 15:38, Bill Gunshannon wrote:
>> In article<i8pms6$v0t$1@iltempo.update.uu.se>,
>> 	Johnny Billquist<bqt@softjar.se>  writes:
>>> On 2010-10-08 23:23, JF Mezei wrote:
>>>> Bill Gunshannon wrote:
>>>>
>>>>> Bullcrap.  OS-X is the Mach Kernel with the BSD Userland (which also
>>>>> contains a lot of GNU) running on top of it.  The only Apple contribution
>>>>> is their wierd window manager.
>>>>
>>>> Apple did not just "use" Mach. It fixed it. There were good reason why
>>>> it had not been widely adopted in the past. Apple fixed those reasons.
>>>> (I rememer reading an article about this, but the technical aspect elude
>>>> me at the moment).
>>>
>>> Well, technically, OS-X isn't much Mach at all. It's mostly NextSTEP
>>> with FreeBSD/NetBSD thrown in. NextSTEP in turn was based on Mach,
>>> people say, but that is in a very limited way, as NextSTEP was also BSD
>>> based, with just a few cherry-picked Mach:isms thrown in.
>>> A serious Mach lover would not recognize himself in OS-X.
>>>
>>> I know that everyone says that OS-X is Mach based, but I'd like anyone
>>> to show me where. The Mach kernel was based around a few simple
>>> concepts. It's a microkernel, with an extensive IPC having features and
>>> possibilities not existing in more classical Unixes. Neither of these
>>> two features have been adopted by anything else.
>>> The third part is the VM system, which you might argue all modern BSD
>>> based systems use, which was adopted from Mach.
>>>
>>> So, no. OS-X is definitely not the Mach kernel with the BSD Userland.
>>
>> OK, I'll concede that although it really appears to depend on where you
>> read.
> 
>:-)
> If you ever take a step back when using a MAC, you'll easily see that 
> it's not Mach. It is not a micro kernel, but follows the traditional 
> Unix monolithic kernel philosophy. It also do IPC the same way as any 
> other Unix, and don't use message passing for everything (including 
> system calls) like Mach do (did).
> 
> Mach was for a while very hip and cool to associate yourself with, even 
> though noone really did adopt most of it.
> 
> Nowadays, Mach is pretty dead.
> 
>>> It's mostly BSD kernel with BSD userland.
>>
>> Which would make it even more just another distro of Unix.
> 
> Indeed! And it is... The only Apple-specific thing is the GUI, and the 
> semi-"get-rid-of text-readable-configuration-files". Apple in all their 
> wisdom implemented a hellish binary configuration data structure 
> instead, so that you need to use their tools to change a lot of things, 
> instead of just poking in some nice text files.
> 
>>>                                             If you can find anything of
>>> Mach in there, please let me know.
>>>
>>> There is quite a lot of information in Wikipedia about Mach, OS-X,
>>> NextSTEP and XNU, for anyone to read if they are interested.
>>
>> I am  not inclined to tear it appart to see what's at the guts.  It's
>> a religion anyway and I am not likely to conbince anyone anyway.
> 
>:-)
> I don't think we need either. I believe your original claim was that OS 
> X was just something Apple grabbed from somewhere else and repackaged. I 
> think we can all agree on this, just as we can agree that it's just 
> another Unix...
> 
>>> That said, someone else claimed that Linux was bigger than OS-X. That
>>> don't seem to be true. OS-X as a Unix system overshadows all other Unix
>>> variants combined, when it comes to deployed systems (yes, add all
>>> Linux, Solaris, xBSD, OSF/1, Ultrix, whatever together, and it's fewer
>>> systems than the OS-X ones...)
>>
>> Well, I assume you are getting these numbers by counting iPhones and
>> other devices with embeded OS-X.  If that's the case yopu really need
>> to count all the systems with embedded versions of the other OSes as
>> well.
> 
> No, I was just talking about normal computers. The most obvious number 
> being web clients, which is very popular to collect statistics on. And 
> those outnumber servers by such a wide margin that servers don't show on 
> the radar anyway.
> 
> If we start counting embedded systems, I think the numbers might look 
> different, but I'm not sure they are different enough to be interesting.
> 
> I know of a few systems where Linux is embedded, but I have not seen a 
> single one where production are counted in the millions.
> 
> But I'm sure you can find some, if you search. However, most embedded 
> systems run more dedicated OSes.
> 
>> How many Netgear switches have been sold world-wide?
>> How many Linksys Access Points?
>> How many Linksys/Cisco NSS boxes?
>>
>> All of them have Linux under the hood.  And there are hundreds more
>> each having sold thousands of units.
>>
>> The original argument was about desktop usage as opposed to server
>> usage and wasn't really about embeded systems at all.
> 
> Yeah. And desktop Linux is very small compared to OS X. But on the other 
> hand, in my eyes, they are all Unix, and are all in a small minority to 
> Windows. :-)
> 
> See http://en.wikipedia.org/wiki/Usage_share_of_operating_systems for 
> some numbers. But to make it short:
> 
> Windows XP (48.32%)
> Windows 7 (19.81%)
> Windows Vista (18.43%)
> Mac OS X (6.42%)
> iOS (iPhone) (1.40%)
> Linux (1.34%)
> 
> Meaning that iPhone alone is more common than Linux, and OS X is more 
> than 4 times as common as Linux.
> 

And I would really need to see some verification of the source of those
numbers before I would be likely to believe them.  Especially considering
the listed source.  For example, how do you collect the numbers for Linux
systems when there is very little sales record and no form of registration?

I have no doubt about Windows holding the lion's share but I seriously
doubt that OS-X outnumbers all different distros of Linux.  I am not
even sure I would accept that they outnumber all other BSD systems.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999@cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>   

In article <8hega0FvkjU16@mid.individual.net>, Bob Eager <rde42@spamcop.net> writes:
>On Sun, 10 Oct 2010 18:27:23 +0000, VAXman- wrote:
>
>> In article <8he7mnFr4lU2@mid.individual.net>, billg999@cs.uofs.edu (Bill
>> Gunshannon) writes:
>>>In article <00AA4BE3.E1850FC7@sendspamhere.org>,
>>>	VAXman-  @SendSpamHere.ORG writes:
>>>> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>, "John
>>>> Wallace" <not@any.price> writes:
>>>>>
>>>>>Wrt access points: Linksys used to make a famous Access Point with
>>>>>Linux under the hood (the WRT54G), and it was popular target for
>>>>>people who wanted a hardware platform for a Linux-based whatever.
>>>>>Others have done Linux in this kind of kit too, in their DSL routers
>>>>>and so on.
>>>>>
>>>>>
>>>> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or
>>>> twice or even thrice a day, it would need to be rebooted as it froze
>>>> up.  Reboot, of course, meaning a power cycle because its web based
>>>> controls were inacess- ible when the units froze.  I'm still suffering
>>>> one.  I need to find a new "access point" (not wireless router) one of
>>>> these days.
>>>
>>>You seem to have problems with everythng you use.  I, on the other hand,
>>>have a network of WAP54G's running through the building from our
>>>department are to the classroom area so that faculty can just pickup
>>>their laptops and walk between their offices and the classrooms while
>>>maintaining connectivity.  System has been in place since the initial
>>>release of the WAP54G (I really don't remember when we first installed
>>>them but it was at least 4 years ago). The only time they get rebooted
>>>is when the power in the building fails and since the introduction of
>>>POE adapters they have run off one power connection with a UPS so that
>>>power failures are non- existant.  They never lockup and I never reboot
>>>them.  They have run like this thru 2 6ix-month periods when I was away
>>>witht he Army without a single failure.  Are you sure you are not really
>>>Joe Btfsplk?
>> 
>> Here's the firmware rev.
>> 
>> Firmware Version:  	v3.04, Dec 27, 2007
>> 
>> Perhaps, you have the version of the code that works?  I'd love to know.
>> The rest of the networking gear here is Cisco.  I'd just like a WAP that
>> is reliable.  Presently, it is so UNRELIABLE that it is plugged into one
>> of my APC MasterSwitch outlets so that it can SNMP power-cycled
>> remotely.
>> 
>> FWIW, the firmware site for WAP54G is in .NZ and is NO LONGER
>> accessible.
>> 
>> There's reportedly firmware update V3.1 and a V3.04.03 version.
>
>Sigh. We have to help you out again...this URL was found VIA linksys.com, 
>and was obvious. Unless .com has moved to New Zealand...
>
> http://homesupport.cisco.com/en-us/wireless/lbc/WAP54G

I was there before.  Clicking on:

   Update WAP54G Firmware

   Before you download the firmware for your WAP54G you need to be 
   sure whats version of WAP54G you are...

Great use of the English language, BTW.  This takes me to the links which
have the firmware at .NZ URLs which are no longer accessible.

-- 
VAXman- A Bored Certified VMS Kernel Mode Hacker    VAXman(at)TMESIS(dot)ORG

All your spirit rack abuses, come to haunt you back by day.
All your Byzantine excuses, given time, given you away.

One needs to carefully define what is a Linux/Unix system.

An embedded system which gives you no access to a command line with unix
commands probably shouldn't be considered a "Unix" system.


But what happens when such a device can be hacked to grant access to the
command line ?


> bike $  ssh root@10.0.0.158
> root@10.0.0.158's password: chocolate


> Executing /private/etc/profile.d/coreutils.sh
> JF-Mezei-iPhone:~ root# ls
> :2eDS_Store  Library  Media  Network Trash Folder  Temporary Items

> JF-Mezei-iPhone:~ root# cd /

> JF-Mezei-iPhone:/ root# ls

> :2eTemporaryItems  Developer  Network Trash Folder  Temporary Items  bin   cores  etc  mnt      sbin  usr
> Applications       Library    System                User             boot  dev    lib  private  tmp   var

> JF-Mezei-iPhone:/ root# 

Although the jailbroken iphone doesn't come with ALL unix utilities, it
is functionally unix at the command line. Many utilities have been
ported to the iPhone.

Until Apple enables the "Terminal.App" to grant access to the command
line and the file system, I would say that iPhones (and iPad, iPodTouch)
are not Unix devices.

But a jailbroken unit probably is a Unix device because you get access
to the Unix command line and functionality.


Similarly, a vanilla WRTG54GL router may run Linux under the hood, but
users don't see it. However, those who change the firmware to Tomato or
another flavour would get access to command line.

Perhaps OS market share should be broken into "devices" and "general
purpose computers".

re: null terminated strings.


My theory is that back in the early days of Unix, most data structures
were fixed length in nature. You read records from some file with fixed
field lengths into fixed lengths variables. The null terminated string
was not too much a problem back then.

By the time people started to dab into variable sized structures (such
as a text field cabable of containing pagragraphs returned by a GUI
framework) it the null terminated string was far too entrenched to be
changed. And problems occur when you deal with variable size fields that
can have a large range of length,

On Sun, 10 Oct 2010 22:02:13 +0000, VAXman- wrote:

> In article <8hega0FvkjU16@mid.individual.net>, Bob Eager
> <rde42@spamcop.net> writes:
>>On Sun, 10 Oct 2010 18:27:23 +0000, VAXman- wrote:
>>
>>> In article <8he7mnFr4lU2@mid.individual.net>, billg999@cs.uofs.edu
>>> (Bill Gunshannon) writes:
>>>>In article <00AA4BE3.E1850FC7@sendspamhere.org>,
>>>>	VAXman-  @SendSpamHere.ORG writes:
>>>>> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>,
>>>>> "John Wallace" <not@any.price> writes:
>>>>>>
>>>>>>Wrt access points: Linksys used to make a famous Access Point with
>>>>>>Linux under the hood (the WRT54G), and it was popular target for
>>>>>>people who wanted a hardware platform for a Linux-based whatever.
>>>>>>Others have done Linux in this kind of kit too, in their DSL routers
>>>>>>and so on.
>>>>>>
>>>>>>
>>>>> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or
>>>>> twice or even thrice a day, it would need to be rebooted as it froze
>>>>> up.  Reboot, of course, meaning a power cycle because its web based
>>>>> controls were inacess- ible when the units froze.  I'm still
>>>>> suffering one.  I need to find a new "access point" (not wireless
>>>>> router) one of these days.
>>>>
>>>>You seem to have problems with everythng you use.  I, on the other
>>>>hand, have a network of WAP54G's running through the building from our
>>>>department are to the classroom area so that faculty can just pickup
>>>>their laptops and walk between their offices and the classrooms while
>>>>maintaining connectivity.  System has been in place since the initial
>>>>release of the WAP54G (I really don't remember when we first installed
>>>>them but it was at least 4 years ago). The only time they get rebooted
>>>>is when the power in the building fails and since the introduction of
>>>>POE adapters they have run off one power connection with a UPS so that
>>>>power failures are non- existant.  They never lockup and I never
>>>>reboot them.  They have run like this thru 2 6ix-month periods when I
>>>>was away witht he Army without a single failure.  Are you sure you are
>>>>not really Joe Btfsplk?
>>> 
>>> Here's the firmware rev.
>>> 
>>> Firmware Version:  	v3.04, Dec 27, 2007
>>> 
>>> Perhaps, you have the version of the code that works?  I'd love to
>>> know. The rest of the networking gear here is Cisco.  I'd just like a
>>> WAP that is reliable.  Presently, it is so UNRELIABLE that it is
>>> plugged into one of my APC MasterSwitch outlets so that it can SNMP
>>> power-cycled remotely.
>>> 
>>> FWIW, the firmware site for WAP54G is in .NZ and is NO LONGER
>>> accessible.
>>> 
>>> There's reportedly firmware update V3.1 and a V3.04.03 version.
>>
>>Sigh. We have to help you out again...this URL was found VIA
>>linksys.com, and was obvious. Unless .com has moved to New Zealand...
>>
>> http://homesupport.cisco.com/en-us/wireless/lbc/WAP54G
> 
> I was there before.  Clicking on:
> 
>    Update WAP54G Firmware
> 
>    Before you download the firmware for your WAP54G you need to be sure
>    whats version of WAP54G you are...
> 
> Great use of the English language, BTW.  This takes me to the links
> which have the firmware at .NZ URLs which are no longer accessible.

That's because you stayed on the 'Overview' tab and started reading the 
forums (notice the word 'Forums' in black, right under 'Overview'?). That 
link was for an unofficial download. The bad English was from someone 
writing on the forum, and pointed to an unofficial copy of 3.04.03 in New 
Zealand.

To do it *right*, go to the page I linked to. Select the 'Downloads' tab 
(surprise, surprise), instead of trying to read the forums.

Now select your hardware version in the dropdown (3 in your case, I 
guess). The 'Download' links appear in blue. They are not in New Zealand.

Oh well.

-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

JF Mezei <jfmezei.spamnot@vaxination.ca> wrote:
> One needs to carefully define what is a Linux/Unix system.
 
> An embedded system which gives you no access to a command line with unix
> commands probably shouldn't be considered a "Unix" system.
 
> But what happens when such a device can be hacked to grant 
> access to the command line ?

There is a network device that allows one to enter a ping address
on a web page.  If you enter, for example,

127.0.0.1 && ls

It will show the ping result followed by a list of files.
That is, it executes

ping 127.0.0.1 && ls

as a unix command line.

-- glen

On Sun, 10 Oct 2010 18:11:44 -0400, JF Mezei wrote:

> re: null terminated strings.
> 
> My theory is that back in the early days of Unix, most data structures
> were fixed length in nature. You read records from some file with fixed
> field lengths into fixed lengths variables. The null terminated string
> was not too much a problem back then.

Afraid I don't go along with that, although it's a tempting theory. The 
whole ethos of UNIX is that nothing is record-based, and little is fixed  
length - delimiters tend to be used to delimit fields, rather than their 
position.

As I recall, the designers didn't want to stick with the BCPL model of 
using one prefix byte for expressing the length of a string. The main 
reason was that it limited strings to a maximum of 255 bytes (usually), 
although that would vary on (say) a PDP-10, where C was planned to use 9 
bits for each character (they wanted characters to divide easily into 
words). Use of varying 'size' fields would have affected portability. Of 
course, they could have used two bytes for the length, but that would 
have been wasteful in most cases (and this was a basic PDP-11 without 
split I/D space). It would also have thrown up portability issues 
relating to endian-ness (even between the PDP-11 (little endian) and 
PDP-10, where (as I recall) byte pointers worked from left to right 
(effectively big-endian)).

-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

In article <8hetbgFvkjU20@mid.individual.net>, Bob Eager <rde42@spamcop.net> writes:
>On Sun, 10 Oct 2010 22:02:13 +0000, VAXman- wrote:
>
>> In article <8hega0FvkjU16@mid.individual.net>, Bob Eager
>> <rde42@spamcop.net> writes:
>>>On Sun, 10 Oct 2010 18:27:23 +0000, VAXman- wrote:
>>>
>>>> In article <8he7mnFr4lU2@mid.individual.net>, billg999@cs.uofs.edu
>>>> (Bill Gunshannon) writes:
>>>>>In article <00AA4BE3.E1850FC7@sendspamhere.org>,
>>>>>	VAXman-  @SendSpamHere.ORG writes:
>>>>>> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>,
>>>>>> "John Wallace" <not@any.price> writes:
>>>>>>>
>>>>>>>Wrt access points: Linksys used to make a famous Access Point with
>>>>>>>Linux under the hood (the WRT54G), and it was popular target for
>>>>>>>people who wanted a hardware platform for a Linux-based whatever.
>>>>>>>Others have done Linux in this kind of kit too, in their DSL routers
>>>>>>>and so on.
>>>>>>>
>>>>>>>
>>>>>> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or
>>>>>> twice or even thrice a day, it would need to be rebooted as it froze
>>>>>> up.  Reboot, of course, meaning a power cycle because its web based
>>>>>> controls were inacess- ible when the units froze.  I'm still
>>>>>> suffering one.  I need to find a new "access point" (not wireless
>>>>>> router) one of these days.
>>>>>
>>>>>You seem to have problems with everythng you use.  I, on the other
>>>>>hand, have a network of WAP54G's running through the building from our
>>>>>department are to the classroom area so that faculty can just pickup
>>>>>their laptops and walk between their offices and the classrooms while
>>>>>maintaining connectivity.  System has been in place since the initial
>>>>>release of the WAP54G (I really don't remember when we first installed
>>>>>them but it was at least 4 years ago). The only time they get rebooted
>>>>>is when the power in the building fails and since the introduction of
>>>>>POE adapters they have run off one power connection with a UPS so that
>>>>>power failures are non- existant.  They never lockup and I never
>>>>>reboot them.  They have run like this thru 2 6ix-month periods when I
>>>>>was away witht he Army without a single failure.  Are you sure you are
>>>>>not really Joe Btfsplk?
>>>> 
>>>> Here's the firmware rev.
>>>> 
>>>> Firmware Version:  	v3.04, Dec 27, 2007
>>>> 
>>>> Perhaps, you have the version of the code that works?  I'd love to
>>>> know. The rest of the networking gear here is Cisco.  I'd just like a
>>>> WAP that is reliable.  Presently, it is so UNRELIABLE that it is
>>>> plugged into one of my APC MasterSwitch outlets so that it can SNMP
>>>> power-cycled remotely.
>>>> 
>>>> FWIW, the firmware site for WAP54G is in .NZ and is NO LONGER
>>>> accessible.
>>>> 
>>>> There's reportedly firmware update V3.1 and a V3.04.03 version.
>>>
>>>Sigh. We have to help you out again...this URL was found VIA
>>>linksys.com, and was obvious. Unless .com has moved to New Zealand...
>>>
>>> http://homesupport.cisco.com/en-us/wireless/lbc/WAP54G
>> 
>> I was there before.  Clicking on:
>> 
>>    Update WAP54G Firmware
>> 
>>    Before you download the firmware for your WAP54G you need to be sure
>>    whats version of WAP54G you are...
>> 
>> Great use of the English language, BTW.  This takes me to the links
>> which have the firmware at .NZ URLs which are no longer accessible.
>
>That's because you stayed on the 'Overview' tab and started reading the 
>forums (notice the word 'Forums' in black, right under 'Overview'?). That 
>link was for an unofficial download. The bad English was from someone 
>writing on the forum, and pointed to an unofficial copy of 3.04.03 in New 
>Zealand.
>
>To do it *right*, go to the page I linked to. Select the 'Downloads' tab 
>(surprise, surprise), instead of trying to read the forums.
>
>Now select your hardware version in the dropdown (3 in your case, I 
>guess). The 'Download' links appear in blue. They are not in New Zealand.

That link is the version I now have.  I thought you were pointing me to a
newer version and the Forum with the .NL links hint at newer releases.  I
don't think reinstalling the same version I've already installed is going
to make a huge difference.  It's not Micro$haft... or is it?

-- 
VAXman- A Bored Certified VMS Kernel Mode Hacker    VAXman(at)TMESIS(dot)ORG

All your spirit rack abuses, come to haunt you back by day.
All your Byzantine excuses, given time, given you away.

On Sun, 10 Oct 2010 22:54:54 +0000, VAXman- wrote:

> In article <8hetbgFvkjU20@mid.individual.net>, Bob Eager
> <rde42@spamcop.net> writes:
>>On Sun, 10 Oct 2010 22:02:13 +0000, VAXman- wrote:
>>
>>> In article <8hega0FvkjU16@mid.individual.net>, Bob Eager
>>> <rde42@spamcop.net> writes:
>>>>On Sun, 10 Oct 2010 18:27:23 +0000, VAXman- wrote:
>>>>
>>>>> In article <8he7mnFr4lU2@mid.individual.net>, billg999@cs.uofs.edu
>>>>> (Bill Gunshannon) writes:
>>>>>>In article <00AA4BE3.E1850FC7@sendspamhere.org>,
>>>>>>	VAXman-  @SendSpamHere.ORG writes:
>>>>>>> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>,
>>>>>>> "John Wallace" <not@any.price> writes:
>>>>>>>>
>>>>>>>>Wrt access points: Linksys used to make a famous Access Point with
>>>>>>>>Linux under the hood (the WRT54G), and it was popular target for
>>>>>>>>people who wanted a hardware platform for a Linux-based whatever.
>>>>>>>>Others have done Linux in this kind of kit too, in their DSL
>>>>>>>>routers and so on.
>>>>>>>>
>>>>>>>>
>>>>>>> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or
>>>>>>> twice or even thrice a day, it would need to be rebooted as it
>>>>>>> froze up.  Reboot, of course, meaning a power cycle because its
>>>>>>> web based controls were inacess- ible when the units froze.  I'm
>>>>>>> still suffering one.  I need to find a new "access point" (not
>>>>>>> wireless router) one of these days.
>>>>>>
>>>>>>You seem to have problems with everythng you use.  I, on the other
>>>>>>hand, have a network of WAP54G's running through the building from
>>>>>>our department are to the classroom area so that faculty can just
>>>>>>pickup their laptops and walk between their offices and the
>>>>>>classrooms while maintaining connectivity.  System has been in place
>>>>>>since the initial release of the WAP54G (I really don't remember
>>>>>>when we first installed them but it was at least 4 years ago). The
>>>>>>only time they get rebooted is when the power in the building fails
>>>>>>and since the introduction of POE adapters they have run off one
>>>>>>power connection with a UPS so that power failures are non-
>>>>>>existant.  They never lockup and I never reboot them.  They have run
>>>>>>like this thru 2 6ix-month periods when I was away witht he Army
>>>>>>without a single failure.  Are you sure you are not really Joe
>>>>>>Btfsplk?
>>>>> 
>>>>> Here's the firmware rev.
>>>>> 
>>>>> Firmware Version:  	v3.04, Dec 27, 2007
>>>>> 
>>>>> Perhaps, you have the version of the code that works?  I'd love to
>>>>> know. The rest of the networking gear here is Cisco.  I'd just like
>>>>> a WAP that is reliable.  Presently, it is so UNRELIABLE that it is
>>>>> plugged into one of my APC MasterSwitch outlets so that it can SNMP
>>>>> power-cycled remotely.
>>>>> 
>>>>> FWIW, the firmware site for WAP54G is in .NZ and is NO LONGER
>>>>> accessible.
>>>>> 
>>>>> There's reportedly firmware update V3.1 and a V3.04.03 version.
>>>>
>>>>Sigh. We have to help you out again...this URL was found VIA
>>>>linksys.com, and was obvious. Unless .com has moved to New Zealand...
>>>>
>>>> http://homesupport.cisco.com/en-us/wireless/lbc/WAP54G
>>> 
>>> I was there before.  Clicking on:
>>> 
>>>    Update WAP54G Firmware
>>> 
>>>    Before you download the firmware for your WAP54G you need to be
>>>    sure whats version of WAP54G you are...
>>> 
>>> Great use of the English language, BTW.  This takes me to the links
>>> which have the firmware at .NZ URLs which are no longer accessible.
>>
>>That's because you stayed on the 'Overview' tab and started reading the
>>forums (notice the word 'Forums' in black, right under 'Overview'?).
>>That link was for an unofficial download. The bad English was from
>>someone writing on the forum, and pointed to an unofficial copy of
>>3.04.03 in New Zealand.
>>
>>To do it *right*, go to the page I linked to. Select the 'Downloads' tab
>>(surprise, surprise), instead of trying to read the forums.
>>
>>Now select your hardware version in the dropdown (3 in your case, I
>>guess). The 'Download' links appear in blue. They are not in New
>>Zealand.
> 
> That link is the version I now have.  I thought you were pointing me to
> a newer version and the Forum with the .NL links hint at newer releases.
>  I don't think reinstalling the same version I've already installed is
> going to make a huge difference.  It's not Micro$haft... or is it?

You said you had version 3.04. The one I pointed you to was 3.04.03. Now, 
if *that* is what you have, you should have said so! The NZ one you were 
complaining about not getting is also 3.04.03, so I assumed you hadn't 
got that one.

-- 
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

In article <4cb23a21$0$19545$c3e8da3$2e0018d8@news.astraweb.com>, JF Mezei
<jfmezei.spamnot@vaxination.ca> writes:
> re: null terminated strings.
> 
> 
> My theory is that back in the early days of Unix, most data structures
> were fixed length in nature. You read records

Unix has no records, just sacks of bytes.

> from some file with fixed
> field lengths into fixed lengths variables. 
> The null terminated string
> was not too much a problem back then.

It isn't much of a problem today either.
Just learn how to use malloc() & friends.
 

In article <8hev3bFvkjU23@mid.individual.net>, Bob Eager <rde42@spamcop.net> writes:
>On Sun, 10 Oct 2010 22:54:54 +0000, VAXman- wrote:
>
>> In article <8hetbgFvkjU20@mid.individual.net>, Bob Eager
>> <rde42@spamcop.net> writes:
>>>On Sun, 10 Oct 2010 22:02:13 +0000, VAXman- wrote:
>>>
>>>> In article <8hega0FvkjU16@mid.individual.net>, Bob Eager
>>>> <rde42@spamcop.net> writes:
>>>>>On Sun, 10 Oct 2010 18:27:23 +0000, VAXman- wrote:
>>>>>
>>>>>> In article <8he7mnFr4lU2@mid.individual.net>, billg999@cs.uofs.edu
>>>>>> (Bill Gunshannon) writes:
>>>>>>>In article <00AA4BE3.E1850FC7@sendspamhere.org>,
>>>>>>>	VAXman-  @SendSpamHere.ORG writes:
>>>>>>>> In article <WemdneTCf4v1LizRnZ2dnUVZ8nidnZ2d@brightview.co.uk>,
>>>>>>>> "John Wallace" <not@any.price> writes:
>>>>>>>>>
>>>>>>>>>Wrt access points: Linksys used to make a famous Access Point with
>>>>>>>>>Linux under the hood (the WRT54G), and it was popular target for
>>>>>>>>>people who wanted a hardware platform for a Linux-based whatever.
>>>>>>>>>Others have done Linux in this kind of kit too, in their DSL
>>>>>>>>>routers and so on.
>>>>>>>>>
>>>>>>>>>
>>>>>>>> Linksys made an "infamous" Access Point too, the WAP54G.  Once, or
>>>>>>>> twice or even thrice a day, it would need to be rebooted as it
>>>>>>>> froze up.  Reboot, of course, meaning a power cycle because its
>>>>>>>> web based controls were inacess- ible when the units froze.  I'm
>>>>>>>> still suffering one.  I need to find a new "access point" (not
>>>>>>>> wireless router) one of these days.
>>>>>>>
>>>>>>>You seem to have problems with everythng you use.  I, on the other
>>>>>>>hand, have a network of WAP54G's running through the building from
>>>>>>>our department are to the classroom area so that faculty can just
>>>>>>>pickup their laptops and walk between their offices and the
>>>>>>>classrooms while maintaining connectivity.  System has been in place
>>>>>>>since the initial release of the WAP54G (I really don't remember
>>>>>>>when we first installed them but it was at least 4 years ago). The
>>>>>>>only time they get rebooted is when the power in the building fails
>>>>>>>and since the introduction of POE adapters they have run off one
>>>>>>>power connection with a UPS so that power failures are non-
>>>>>>>existant.  They never lockup and I never reboot them.  They have run
>>>>>>>like this thru 2 6ix-month periods when I was away witht he Army
>>>>>>>without a single failure.  Are you sure you are not really Joe
>>>>>>>Btfsplk?
>>>>>> 
>>>>>> Here's the firmware rev.
>>>>>> 
>>>>>> Firmware Version:  	v3.04, Dec 27, 2007
>>>>>> 
>>>>>> Perhaps, you have the version of the code that works?  I'd love to
>>>>>> know. The rest of the networking gear here is Cisco.  I'd just like
>>>>>> a WAP that is reliable.  Presently, it is so UNRELIABLE that it is
>>>>>> plugged into one of my APC MasterSwitch outlets so that it can SNMP
>>>>>> power-cycled remotely.
>>>>>> 
>>>>>> FWIW, the firmware site for WAP54G is in .NZ and is NO LONGER
>>>>>> accessible.
>>>>>> 
>>>>>> There's reportedly firmware update V3.1 and a V3.04.03 version.
>>>>>
>>>>>Sigh. We have to help you out again...this URL was found VIA
>>>>>linksys.com, and was obvious. Unless .com has moved to New Zealand...
>>>>>
>>>>> http://homesupport.cisco.com/en-us/wireless/lbc/WAP54G
>>>> 
>>>> I was there before.  Clicking on:
>>>> 
>>>>    Update WAP54G Firmware
>>>> 
>>>>    Before you download the firmware for your WAP54G you need to be
>>>>    sure whats version of WAP54G you are...
>>>> 
>>>> Great use of the English language, BTW.  This takes me to the links
>>>> which have the firmware at .NZ URLs which are no longer accessible.
>>>
>>>That's because you stayed on the 'Overview' tab and started reading the
>>>forums (notice the word 'Forums' in black, right under 'Overview'?).
>>>That link was for an unofficial download. The bad English was from
>>>someone writing on the forum, and pointed to an unofficial copy of
>>>3.04.03 in New Zealand.
>>>
>>>To do it *right*, go to the page I linked to. Select the 'Downloads' tab
>>>(surprise, surprise), instead of trying to read the forums.
>>>
>>>Now select your hardware version in the dropdown (3 in your case, I
>>>guess). The 'Download' links appear in blue. They are not in New
>>>Zealand.
>> 
>> That link is the version I now have.  I thought you were pointing me to
>> a newer version and the Forum with the .NL links hint at newer releases.
>>  I don't think reinstalling the same version I've already installed is
>> going to make a huge difference.  It's not Micro$haft... or is it?
>
>You said you had version 3.04. The one I pointed you to was 3.04.03. Now, 
>if *that* is what you have, you should have said so! The NZ one you were 
>complaining about not getting is also 3.04.03, so I assumed you hadn't 
>got that one.

I don't believe that the firmware shows the .03 but it does show the date.
The date I posted and that link share the same.

There was a V3.1 mentioned in the forum.


-- 
VAXman- A Bored Certified VMS Kernel Mode Hacker    VAXman(at)TMESIS(dot)ORG

All your spirit rack abuses, come to haunt you back by day.
All your Byzantine excuses, given time, given you away.

Michael Kraemer wrote:

>> My theory is that back in the early days of Unix, most data structures
>> were fixed length in nature. You read records
> 
> Unix has no records, just sacks of bytes.

Applications in the 1960s were record based. Remember punched cards ?
Max 80 characters.

Just because the OS didn't provide a record management service doesn't
mean that applications weren't record based. printers were 132 columns
max, so applications had to fit each output line in 132 columns plus the
trailing linefeed carriage return.

There was far less of a need for dynamic allocation back then.

So the disadvantages/dangers of non-counted strings were not felt
strongly in such applications,  and avoiding a 2 byte counted string
provide space saving.

Also, one would have had to provide 4 bytes. Two bytes  for allocated
size and 2 bytes for actual string size.