Mailer flag 'o'

Hi all,

I'm trying to get sendmail to deliver to Dovecot LDA ('deliver').

It works like this:

FEATURE(`local_procmail',`/usr/local/libexec/dovecot/deliver',
   `deliver -d $u')dnl

But I would like to use the 'w' and 'o' flags, so that deliver is invoked 
as the user who owns the mailbox, like this:

FEATURE(`local_procmail',`/usr/local/libexec/dovecot/deliver',
   `deliver',`Pfhnwo9')dnl

SPfhn9 are the default flags for local_procmail, so I have just added 'w' 
and 'o', and removed 'S'.

The Installation and Operation Guide states describes these as:

w	The user must have a valid account on this machine, i.e.,
	getpwnam must succeed. If not, the mail is bounced. See
	also the MailBoxDatabase option. This is required to get
	“.forward” capability.

o	Always run as the owner of the recipient mailbox. Normally
	sendmail runs as the sender for locally generated mail or as
	“daemon” (actually, the user specified in the u option) when
	delivering network mail. The normal behavior is required by most
	local mailers, which will not allow the envelope sender address
	to be set unless the mailer is running as daemon. This flag is
	ignored if the S flag is set.

S	Don’t reset the userid before calling the mailer. This would be
	used in a secure environment where sendmail ran as root. This
	could be used to avoid forged addresses. If the U= field is also
	specified, this flag causes the effective user id to be set to
	that user.

However, when I try to run a message with these settings, sendmail 
doesn't work.  If I run 

root@classic.viza.gotdns.com # sendmail -v -qf

Running /var/spool/mqueue/m5PGUgkI010008 (sequence 1 of 1)
<tcv@classic.viza.gotdns.com>... Connecting to local...
<tcv@classic.viza.gotdns.com>... Deferred: Connection timed out with 
classic.viza.gotdns.com

It blocked at the "Connecting to local..." for several minutes.

By replacing deliver with a tiny program that logs its uid, euid and what 
it can read, I can tell that deliver is being started properly with the 
uid and euid of the destination user (tcv above), and that read
( STDIN_FILENO, ..) returns 0.

By making the tiny program do close( STDOUT_FILENO ) before trying to 
read, sendmail does not block:

Running /var/spool/mqueue/m5PGUgkI010008 (sequence 1 of 1)
<tcv@classic.viza.gotdns.com>... Connecting to local...
<tcv@classic.viza.gotdns.com>... Deferred: Connection reset by 
classic.viza.gotdns.com

The above is instant.

So why does adding the 'o' option to a mailer make it try to read from 
the standard output of the mailer program?

I guess it is waiting for a "220 host SMTP ready" line or similar, but 
there is nothing in the documentation to say that it does that.  How do I 
get it to run as the correct user but just blindly pipe the message 
through, like FEATURE(`local_procmail') normally does?

Thanks

viza.
0
tom.viza2 (51)
6/25/2008 4:54:54 PM
comp.mail.sendmail 13479 articles. 0 followers. jfretby (35) is leader. Post Follow

10 Replies
175 Views

Similar Articles

[PageSpeed] 22
On 06/25/08 11:54, viza wrote:
Not an answer to your question, but some thoughts about it.

> FEATURE(`local_procmail',`/usr/local/libexec/dovecot/deliver',
>    `deliver',`Pfhnwo9')dnl

Rather than re-defining the binary of the `local_procmail' ""mailer, why 
not create a new ""mailer for Dovecot?  You should be able to create a 
new ""mailer by copying the cf/feature/local_procmail.m4 file to 
something like cf/feature/dovecot.m4 and editing accordingly.  You will 
get the same effect as editing the `local_procmail' ""mailer as far as 
delivery is concerned.  As an added benefit, the `local_procmail' 
""mailer will still be valid so you could chose to use it for other 
things and even better is you could submit the cf/feature/dovecot.m4 
file back to Dovecot for inclusion in to future releases for much easier 
inclusion via a simple (copy a file if it does not exist) 
"FEATURE(`Dovecot', <bla>)" line in your .mc file.

> SPfhn9 are the default flags for local_procmail, so I have just added 'w' 
> and 'o', and removed 'S'.

I don't know what the other flags indicate off the top of my head, and 
as such I'll have to look them up to better respond.

> So why does adding the 'o' option to a mailer make it try to read from 
> the standard output of the mailer program?
> 
> I guess it is waiting for a "220 host SMTP ready" line or similar, but 
> there is nothing in the documentation to say that it does that.  How do I 
> get it to run as the correct user but just blindly pipe the message 
> through, like FEATURE(`local_procmail') normally does?

Normal ""mailers (local, prog, mailman, procmail) send data to STDIN of 
the LDA and may read the exit status of the LDA.  However based on what 
you say, it sounds like your your sendmail is wanting to speak LMTP with 
the ""mailer.



Grant. . . .

0
gtaylor (1357)
6/25/2008 5:30:49 PM
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Wed, 25 Jun 2008 12:30:49 -0500, Grant Taylor wrote:

> On 06/25/08 11:54, viza wrote:
> Not an answer to your question, but some thoughts about it.

>> FEATURE(`local_procmail',`/usr/local/libexec/dovecot/deliver',
>>    `deliver',`Pfhnwo9')dnl

[snip]

> Normal ""mailers (local, prog, mailman, procmail) send data to STDIN of
> the LDA and may read the exit status of the LDA.  However based on what
> you say, it sounds like your your sendmail is wanting to speak LMTP with
> the ""mailer.

I had the same problem. It seems that sendmail tries to speak LMTP if the
third argument to that FEATURE does NOT include $u. Try

FEATURE(`local_procmail',`/usr/local/libexec/dovecot/deliver',
   `deliver $u',`Pfhnwo9')dnl

but I don't know if that extra argument will mess up dovecot/deliver.


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)

iD8DBQFIYo7VL6j7milTFsERAg4KAJwO0PoVDa8/0W6IxRUB8Te6H/zbcACfSOYh
BhbgoD99RZqwcgYXUKTFvMQ=
=mcLn
-----END PGP SIGNATURE-----


0
carl8110 (134)
6/25/2008 6:31:48 PM
Hi,

On Wed, 25 Jun 2008 11:31:48 -0700, Carl Byington wrote:
>> On 06/25/08 11:54, viza wrote:
 
>>> FEATURE(`local_procmail',`/usr/local/libexec/dovecot/deliver',
>>>    `deliver',`Pfhnwo9')dnl
>
> So why does adding the 'o' option to a mailer make it try to read from
> the standard output of the mailer program?
> 
> I had the same problem. It seems that sendmail tries to speak LMTP if
> the third argument to that FEATURE does NOT include $u.

Thanks Carl, that's really useful.  Has this been formally reported in a 
bugzilla or whatever else sendmail uses?

> Try
> FEATURE(`local_procmail',`/usr/local/libexec/dovecot/deliver',
>    `deliver $u',`Pfhnwo9')dnl
> 
> but I don't know if that extra argument will mess up dovecot/deliver.

Spare arguments will upset dovecot deliver.  My first thought was to just 
set the argument vector to just `$u' (after making sure that deliver 
doesn't care what argv[0] is), but that doesn't work with sendmail either.

So, here is a complete workaround using a wrapper: (source code follows)

First compile the wrapper program, defining the location of deliver and 
outputting into the same directory, in my case:

cc -DEXECUTABLE='"/usr/local/libexec/dovecot/deliver"' \
  -o /usr/local/libexec/dovecot/deliver-wrapper deliver-wrapper.c 

and change the ownership and mode of deliver-wrapper to be the same as 
deliver, in my case:

chown root:bin /usr/local/libexec/dovecot/deliver-wrapper
chmod 555 /usr/local/libexec/dovecot/deliver-wrapper

In sendmail.mc, set the executable to the location of the wrapper and the 
argument vector to include -_ $u.  Use whatever flags you need, although 
this workaround isn't necessary if you aren't trying to use the flag 
'o' (because you would be using -d $u in any other case).

FEATURE(`local_procmail',`/usr/local/libexec/dovecot/deliver-wrapper',
   `deliver -_ $u',`Pfhnwo9')dnl

That's it.  Here is the source for the wrapper program:

/* deliver-wrapper.c PUBLIC DOMAIN 2008 tom.viza@gmail.com */
/* remove -_ foo or -_foo from argument vector */

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>

#ifndef EXECUTABLE
#error you must use -DEXECUTABLE='"/usr/bin/whatever"'
#endif

int main( int argc, char **argv ){
  char **dest= argv + 1;
  char **src= dest;

  while( *src ){
    if( '-' == src[0][0] && '_' == src[0][1] ){
      if( src[0][2] )
        ++src;

      else {
        if( src[1] )
          src+= 2;
        else 
          break;        
      }
    }
    else
      *dest++= *src++;
  }
  *dest= NULL;
  execv( EXECUTABLE, argv );  
  perror( "exec " EXECUTABLE );
  exit( EXIT_FAILURE );
}

HTH
viza
0
tom.viza2 (51)
6/25/2008 9:22:04 PM
In article <0Gy8k.63071$7m7.6202@newsfe30.ams2> viza <tom.viza@gmil.com>
writes:
>
>On Wed, 25 Jun 2008 11:31:48 -0700, Carl Byington wrote:
>> 
>> I had the same problem. It seems that sendmail tries to speak LMTP if
>> the third argument to that FEATURE does NOT include $u.
>
>Thanks Carl, that's really useful.  Has this been formally reported in a 
>bugzilla or whatever else sendmail uses?

It's working as designed - from doc/op/op.*:

           If the argument vector does not contain  $u  then
      sendmail will speak SMTP (or LMTP if the mailer flag z
      is specified) to the mailer.

It's pretty reasonable, since there must be some way for sendmail to
tell the mailer program who the recipient(s) is/are... I guess
dovecot/deliver infers that from the uid it is run as, which seems
rather "weird".

[snip wrapper that throws away $u]

Just make sure that you don't have 'm' in the mailer flags - if you do,
sendmail will expand $u to cover all of them for a message with multiple
recipients. It would seem that 'm' is incompatible with 'o', but I'm not
sure what happens if you have both.

--Per Hedeland
per@hedeland.org
0
per71 (2634)
6/25/2008 11:05:25 PM
On Wed, 25 Jun 2008 23:05:25 +0000, Per Hedeland wrote:

> In <0Gy8k.63071$7m7.6202@newsfe30.ams2> viza <tom.viza@gmil.com> writes:
>>On Wed, 25 Jun 2008 11:31:48 -0700, Carl Byington wrote:
>>> 
>>> I had the same problem. It seems that sendmail tries to speak LMTP if
>>> the third argument to that FEATURE does NOT include $u.
>>
>>Thanks Carl, that's really useful.  Has this been formally reported in a
>>bugzilla or whatever else sendmail uses?
> 
> It's working as designed - from doc/op/op.*:
> 
>            If the argument vector does not contain  $u  then
>       sendmail will speak SMTP (or LMTP if the mailer flag z is
>       specified) to the mailer.

That is hideously ambiguous!  I don't know if any policy makers read 
here, but shouldn't there be explicit flags to choose smtp, lmtp or 
piping to stdin?

> It's pretty reasonable, since there must be some way for sendmail to
> tell the mailer program who the recipient(s) is/are... I guess
> dovecot/deliver infers that from the uid it is run as, which seems
> rather "weird".

It's fairly common to discard privileges if they are not required.  Once 
the deliver process is non-privileged, there is only one mailbox it can 
possibly write to!  

Obviously this is only the case in traditional "real accounts" mode, not 
with virtual users, where you wouldn't use this feature.

> Just make sure that you don't have 'm' in the mailer flags - if you do,
> sendmail will expand $u to cover all of them for a message with multiple
> recipients. It would seem that 'm' is incompatible with 'o', but I'm not
> sure what happens if you have both.

Dovecot deliver only does one message at a time anyway.

Thanks

viza
0
tom.viza2 (51)
6/26/2008 10:01:35 AM
In article <3OJ8k.42091$Kb.16669@newsfe29.ams2> viza <tom.viza@gmil.com>
writes:
>On Wed, 25 Jun 2008 23:05:25 +0000, Per Hedeland wrote:
>> 
>> It's working as designed - from doc/op/op.*:
>> 
>>            If the argument vector does not contain  $u  then
>>       sendmail will speak SMTP (or LMTP if the mailer flag z is
>>       specified) to the mailer.
>
>That is hideously ambiguous!

Not at all - it's completely unambiguous, and succinct.	

>  I don't know if any policy makers read 
>here, but shouldn't there be explicit flags to choose smtp, lmtp or 
>piping to stdin?

The choice is not between "piping to stdin" and the others, but between
"recipients on command line" and the others - LMTP is normally on
stdin/stdout. Adding explicit flags would only bring redundancy and
additional possibilities to specify erroneous combinations
(e.g. "SMTP-flag" + $u in Argv).

>It's fairly common to discard privileges if they are not required.  Once 
>the deliver process is non-privileged, there is only one mailbox it can 
>possibly write to!  
>
>Obviously this is only the case in traditional "real accounts" mode, not 
>with virtual users, where you wouldn't use this feature.

Sendmail has no way of knowing how or where the LDA is doing delivery,
thus it's reasonable to require that the mailer definition includes a
way to pass the recipients to the mailer program, and not assume that it
will deduce this from the uid it is running as.

Actually this concept seemed so bizarre that I had to go and look up the
Dovecot docs, and as I expected, "deliver" accepts most all standard LDA
arguments - in particular (from http://wiki.dovecot.org/LDA):

  -d <username>: Destination username.

The "infer recipient from uid" mode seems to be recommended for use in
~/.forward files, where it's perfectly reasonable of course. In a
sendmail mailer definition, it isn't.

>Dovecot deliver only does one message at a time anyway.

The issue was not about multiple messages, but about one message with
multiple recipients. If you are running Dovecot "deliver" in such a way
that it can only determine recipients by checking its uid, there's
obviously no way it can deliver to multiple recipients. This wasn't the
point though - it was that sendmail must be informed about this
limitation (via the absent 'm' flag).

--Per Hedeland
per@hedeland.org
0
per71 (2634)
6/27/2008 12:02:41 PM
Hi

On Fri, 27 Jun 2008 12:02:41 +0000, Per Hedeland wrote:
> In <3OJ8k.42091$Kb.16669@newsfe29.ams2> viza <tom.viza@gmil.com> writes:
>>On Wed, 25 Jun 2008 23:05:25 +0000, Per Hedeland wrote:
>>> 
>>> It's working as designed - from doc/op/op.*:
>>>            If the argument vector does not contain  $u  then
>>>       sendmail will speak SMTP (or LMTP if the mailer flag z is
>>>       specified) to the mailer.
>>
>>That is hideously ambiguous!
> 
> Not at all - it's completely unambiguous, and succinct.

Inferring the format for messages from an almost orthogonal option is 
perhaps not ambiguous, but it does require a new administrator to look in 
places they wouldn't expect to have to for the documentation.

>> but shouldn't there be explicit flags to choose smtp, lmtp or
>>piping to stdin?
> 
> Adding explicit flags would only bring redundancy and additional
> possibilities to specify erroneous combinations (e.g.
> "SMTP-flag" + $u in Argv).

There are dozens of incompatible combinations already!  What if you 
specify 789 in flags? 

If the configuration was explicit then there would be no danger of 
getting it wrong by accident, as it seems both I and Carl Byington did.

>>It's fairly common to discard privileges if they are not required.  Once
>>the deliver process is non-privileged, there is only one mailbox it can
>>possibly write to!
>>
>>Obviously this is only the case in traditional "real accounts" mode, not
>>with virtual users, where you wouldn't use this feature.
> 
> Sendmail has no way of knowing how or where the LDA is doing delivery,
> thus it's reasonable to require that the mailer definition includes a
> way to pass the recipients to the mailer program, and not assume that it
> will deduce this from the uid it is running as.

It IS reasonable to require that the mailer definition includes a
way to pass the recipients to the mailer program.  It IS not reasonable 
to assume that it will deduce this from the uid it is running as, but 
neither is is reasonable to assume that it will use a command line 
argument to do it.  Implicit configuration is a bad idea.

> Actually this concept seemed so bizarre that I had to go and look up the
> Dovecot docs, and as I expected, "deliver" accepts most all standard LDA
> arguments - in particular (from http://wiki.dovecot.org/LDA):
> 
>   -d <username>: Destination username.

It is more complicated than that.  -d 'foo' doesn't mean "deliver to the 
mailbox 'foo'", it means "connect to the authentication agent, lookup the 
string 'foo' and determine if the process has permission to deliver to 
the mailbox that it refers to".

It is usual for normal users to not have access to the authentication 
process, so they must not use the -d option.  Of course, one could 
deliver as root with the -d option, but I'm trying to follow the 
convention of dropping privileges early.
0
tom.viza2 (51)
6/28/2008 10:38:26 AM
In article <Cwo9k.131109$Ek2.97716@newsfe17.ams2> viza
<tom.viza@gmil.com> writes:
>
>On Fri, 27 Jun 2008 12:02:41 +0000, Per Hedeland wrote:
>> In <3OJ8k.42091$Kb.16669@newsfe29.ams2> viza <tom.viza@gmil.com> writes:
>>>On Wed, 25 Jun 2008 23:05:25 +0000, Per Hedeland wrote:
>>>> 
>>>> It's working as designed - from doc/op/op.*:
>>>>            If the argument vector does not contain  $u  then
>>>>       sendmail will speak SMTP (or LMTP if the mailer flag z is
>>>>       specified) to the mailer.
>>>
>>>That is hideously ambiguous!
>> 
>> Not at all - it's completely unambiguous, and succinct.
>
>Inferring the format for messages from an almost orthogonal option is 
>perhaps not ambiguous, but it does require a new administrator to look in 
>places they wouldn't expect to have to for the documentation.
>
>[snip other arguments about the badness of the above]

I can agree that looking at current sendmail, "absence of $u" isn't the
most obvious choice of "SMTP/LMTP selector" - but to get some
understanding of why the choice was made, you need to go back to when it
happened, 25+ years ago when sendmail was first taught to speak SMTP at
all. And given that it hasn't changed since then, it's not exactly
likely to happen now, no matter how upset you get about it.

And it *is* documented - and I would certainly recommend that anyone
writing their own mailer definition *carefully* reads the *entire*
"Define Mailer" section of doc/op/op.*.

>It IS reasonable to require that the mailer definition includes a
>way to pass the recipients to the mailer program.  It IS not reasonable 
>to assume that it will deduce this from the uid it is running as, but 
>neither is is reasonable to assume that it will use a command line 
>argument to do it.  Implicit configuration is a bad idea.

It's not so much assuming as lack of support for the scheme you want -
sendmail knows of two ways to pass the recipient info to the mailer, via
commandline $u or "inline" via SMTP/LMTP - it requires that you do one
or the other. You want a third, "via process uid", and there is no
support for that - so you had to do a little tweaking to make it work.

>> Actually this concept seemed so bizarre that I had to go and look up the
>> Dovecot docs, and as I expected, "deliver" accepts most all standard LDA
>> arguments - in particular (from http://wiki.dovecot.org/LDA):
>> 
>>   -d <username>: Destination username.
>
>It is more complicated than that.  -d 'foo' doesn't mean "deliver to the 
>mailbox 'foo'", it means "connect to the authentication agent, lookup the 
>string 'foo' and determine if the process has permission to deliver to 
>the mailbox that it refers to".

Nevertheless, it seems clear from what I read of the docs that this (as
well as the use of other standard LDA flags such as -f) is how the
authors intend the program to be used as LDA.

>It is usual for normal users to not have access to the authentication 
>process, so they must not use the -d option.

But a mailer definition in sendmail is not "a normal user".

> Of course, one could 
>deliver as root with the -d option, but I'm trying to follow the 
>convention of dropping privileges early.

It's a good goal, but if you end up interfacing sendmail and a LDA such
that both are used in ways that were never intended, it may be reason to
reconsider it.

--Per Hedeland
per@hedeland.org
0
per71 (2634)
6/29/2008 9:14:07 PM
See, http://wiki.dovecot.org/LDA/Sendmail

The recommended way is to use deliver as mailer is SUID.

To trick non-SUID deliver and sendmail, this might work:

/etc/mail/smrsh/dovecot-deliver  -f $u -f $g

so you have $u on command line, but override the value with from.

Bye, ska
0
skg34 (195)
6/30/2008 10:48:02 AM
On Mon, 30 Jun 2008 03:48:02 -0700, ska wrote:

> See, http://wiki.dovecot.org/LDA/Sendmail
> 
> The recommended way is to use deliver as mailer is SUID.

No, that page (remember it's a wikiwiki so could have been written by 
anyone) recommends that when sendmail doesn't run as root.

Dovecot LDA's authors used to state that it should not be installed 
SUID.  I can't find any reference to that now, so maybe they are happy 
that that is now ok.

In any case, the whole title of this thread is "Mailer 'o'" flag, which 
means to run the LDA as the destination user.

> To trick non-SUID deliver and sendmail, this might work:
> 
> /etc/mail/smrsh/dovecot-deliver  -f $u -f $g

That's an interesting idea.  It will work too, until the dovecot 
developers update src/deliver/deliver.c to do more robust argument 
handling.  I'll include it if I get round to documenting the wrapper I 
posted elsewhere in this thread.
0
tom.viza2 (51)
6/30/2008 11:40:33 AM
Reply:
Similar Artilces:

saA Flag
Hi, people at one of our remote site having problem access to exchange server here. they have 506e, 6.3. when I checked their PIX, there are 100s of connections with saA flags, usually i dont see these connection. does anyone know how to fix it? The Tunnel seems to be up. TCP out 192.168.34.146:135 in 192.168.0.194:3387 idle 0:00:01 Bytes 0 flags saA Thanks in advance for any help. In article <423b20ea$1@news.sentex.net>, Rob <rob@hotmail.com> wrote: :people at one of our remote site having problem access to exchange server :here. they have 506e, 6.3. when I checked their PIX, ...

watch that little flag and lets all see who can shoot that target first ! intelligence ratios, mark scores and group dynamics in video
I don;t do games design I design the shelves and the gadgets the gadgets that makers use in compiling their videos...but hey! thats interesting...too! ...

PHP mailer script
I would like to set up a PHP page which my staff can use to select attachments and send them to a specified email address. There will be a list of about 10 alternative attachments (brochures) in PDF format which are stored on the server. These are information brochures for customers, and I dont want any other files to be able to be emailled. So, what is the best approach/scripting approach to do this. Basically I want to have a form with: - Customers name - Customers email address where the email will go to "Customers name <customersemail@isp.com>" - check boxes for t...

File I/O from threads
Hi, I was writing a quick Relay class and ran into a problem. Objects of this class act between two ports and log every TCP/IP that is sent between those ports - according to a specific protocol (in this simple example the first 8 characters contain the length of the data that will follow) These are the important methods of the class. All objects are local to the object and not shared between other objects. def start @server = TCPServer.new(l_h, l_p) @client = TCPSocket.new(r_h, r_p) loop do Thread.start(@server.accept) do |s| # handshake(s) client_t...

asynchronous i/o for hp-ux on oracle?
I've been told that it may be possible to get asynchronous i/o on hp-ux for "cooked" i.e. not raw files on hp-ux 11.31 [11iv3]. Anyone able to confirm that this is the case? strepxe@yahoo.co.uk wrote: > I've been told that it may be possible to get asynchronous i/o on hp-ux > for "cooked" i.e. not raw files on hp-ux 11.31 [11iv3]. Anyone able to > confirm that this is the case? Don't know I have not heard this but am interested in knowing more. You may have to open support requests with both hp and oracle to get more info if you don't have a dif...

distutils compiler flags for extension modules
Hoi, as I got no answers with the previous question (subject: disabling compiler flags in distutils), I thought I should ask the question in a different way: Is there an option to set the compiler flags for a C/C++ extension in distutils? There is the extra_compile_args-option in the Extension class, yet this offers only to give additional flags, but I'd like to have 'total' control about the compile args. Any hint? Thanks Christian On Mar 20, 9:48=A0am, Christian Meesters <meest...@gmx.de> wrote: > as I got no answers with the previous question (subject: disabling...

for sale, o.t.
Hi all, I have the following available for those of you still using these, 4 packs of 5 disks each, teijin brand, mo-128, rewritable 90 mm magneto optical disks, $5.00 a pack of 5 and $4.00 for shipping via media mail, international shipping i can do also but it's gonna cost ya a lot since the rates went sky high. regards, Jerry -- Brought to you from a COMPAQ PROLIANT 3000 running WIN2KPRO, check out www.jerrysmicro.com but don't get too exited! ...

Try to stop thread Using flag
Hi,pythoners: I countered some problems when I try to stop threads using flag. These are my some important codes: ##### mythread.py def run(self): while self.addr != '': ### text waiting for processing if not self.CONTINUE: ### flag for thread ,means to exit the RUN func or not break print self.name post(self.params, self.addr) ### func to process the text self.addr = furl.readline().strip() ###### myapp.py def OnEndProcess(self, event): if self.threadList: for thread in ...

Burst I/O on PPC440GP External Bus
I am trying to perform I/O with a device attached to the external peripheral bus on a PPC440GP embedded processor. I am able to address the device and read and write to it, but I am only able to do non-burst I/O. Do I need to do something special to enable burst transactions? I've programmed the EBC0_B5CR and EBC0_B5AP device control registers appropriately to enable burst mode, but the transactions are still single I/O. Do I need to use PPC-specific instructions to perform burst I/O, or should an ordinary programmed I/O (using 32-bit write operations) be automatically queued up a...

undefined symbol seen in ios.o (libCStd.a)
Hi, While building a shared library , I am getting the following error ----------- Undefined first referenced symbol in file void operator delete[](void*) /opt/SunProd/SUNWspro8/prod/lib/v9/libCstd.a (ios.o) ----------- Any idea on solving this ? Thanks, Mukundan Sorry, can't restrain himself. Do you develop a new Oracle 10g release ? Mukundan Narayanan wrote: > Hi, > > While building a shared library , I am getting the following error > > ----------- > Undefined first referenced > symbol in file > void operator delete[](void*) /opt/SunProd/SUNWspro8/prod/...

US: Programmer/Analyst, Schema/data modeling, w/Business Objects, VB, NSK SQL, o (45292257609)
US: Programmer/Analyst, Schema/data modeling, w/Business Objects, VB, NSK SQL, o (45292257609) ============================================================================================== Position: Programmer/Analyst Reference: SMC01129 Location: Duration: Skills: Functional and technical experience working with relevant software, schema and data modeling, as well as programming languages to include one or more of the following: Business Objects, VB, NSK SQL, C with Health Care data warehouse and decision supp...

MAILER-DAEMON blank from=<>
Greetings to all: I'm not very skilled at searching the archives, but what I've found so far tends to indicate blocking email based on a blank from line is a bad thing to do. However, recently I've noticed that when our server bounces(attempts to return) an email with the "User unknown" message, some sites have refused to accept the blank from line as in: Dec 10 12:28:42 ns1 sendmail[8236]: hBAKSZ7k008236: from=<>, Dec 10 12:28:43 ns1 sendmail[8236]: hBAKSZ7k008236: ruleset=check_rcpt, reject=554 5.7.0 ERROR Probable Forgery. . . Need Valid Return Email Address. ...

Line break
I'm trying to write a two-dimensional array to an ASCII file. The problem is that FORTRAN automatically changes line for every 7th record, instead of writing all the values of one row on one line. How can I fix this? EricT wrote: > I'm trying to write a two-dimensional array to an ASCII file. > The problem is that FORTRAN automatically changes line for every 7th > record, instead of writing all the values of one row on one line. > How can I fix this? Probably you are using list-directed formatting (with the default "*" format), for which the number of numbers p...

"New" flags are permanent on solid state drives
SSDs are generally mounted with the "noatime" option to reduce the number of writes. The problem is, this causes the "N" flags to never get cleared, even when every message in the mbox has been read. Solution? On 2014-03-22, Nomen Nescio wrote: >SSDs are generally mounted with the "noatime" option to reduce the >number of writes. > >The problem is, this causes the "N" flags to never get cleared, even >when every message in the mbox has been read. > >Solution? maildir. * Nomen Nescio on Saturday, March 22, 2014...

Problem with Region flags in US region calibration module
Hi, I have a problem understanding the region flags used in the US region calibration module. I am using these regions to decide areas of the US image where measurements can be drawn. The description for the region flag states that (Bit 0 Transparency ) - "If the region is transparent, then measurements may be done on regions underneath this region." Hence I parse the regions & allow measurements only on regions with Transparency = 0 i.e. it is a opaque region. Is my assumption correct? My question is whether this logic applies only to overlapping regions? If there i...

Application of parity flag
Hi, Please, anybody tell me what are applications of parity flag . Thank you Advait Raut schrieb: > Hi, Please, anybody tell me what are applications of parity flag. Applications that use the parity flag? The parity flag is a bit in the CPU-status-register, wich is set by a result of a arithmetic instruction and can be use for following instructions like jp "offset", to jump when parity flag is set. Dirk On Thu, 21 Oct 2004 10:24:29 +0000 (UTC), Dirk Wolfgang Glomp <spamtrap@crayne.org> wrote: >Advait Raut schrieb: >> Hi, Please, anybody tell me what are...

on closing the DEC rat holes and Borland hacking the applause-o-meter ..
From: jimail Fri Jan 3 07:55:55 1992 To: billg; carls; jeffr; kellyw; maikmap; paulma Cc: steveb Subject: Digital letter Date: Fri: Jan 03 07:54:26 PDT 1992 Mail-Flags: 8000 We are about to send a letter to Digital outlining our conclusions concerning DBC NAS and WOIA. It will be quite direct and concrete with the summary being: 1. We will jointly promote one API set for Windows clients, consisting of Win16, Win32, OLE and Cairo components. ... We do not suggest that we "merge CDA and OLE" or any of the other hundred rat holes that could be entered .. Long term I can't imagin...

Algorithms Notes by Derek O'Connor
Hi Does anyone have the notes written by Prof. Derek O'Connor (of University College Dublin) on Data Structures and Algorithms? I used to read those notes for algos and apparently that site is taken down as he had left. If any one has those notes, please email them to me. Thanks Muralidhar muralidharRemovethiSck@yahoo.com "Murali" <mux@sd.com> writes: > Hi > > Does anyone have the notes written by Prof. Derek O'Connor (of University > College Dublin) on Data Structures and Algorithms? I used to read those > notes for algos and apparently that site...

Jakarta Struts and File I/O
Hi, I've been working on a webapp using Struts. I need to do some file I/O in a few of my Action classes. Is there a way to do this easily? What I mean is, is there an app(or server)-context relative way to read files provided by Struts (or maybe even Tomcat)? A property I can read or an object that does file I/O in a context relative way (ie something like new File(mysteryObj.getServerDir()+"webbapps/my_app/foobar.txt") instead of new File("/all/the/way/from/root/Tomcat 5.0/webapps/my_app/foobar.txt")) that I'm not aware of? The text books I've read don...

cvv sellers dumps tracks 1 and 2 bank logins wu transfers smtp mailer php mailer for sale
Cvv,fullz,banklogins,dumps+pin,track1/2+pin,bank Transfer,wu Transfer,shop admin, etc I AM BOOKING FLIGHT TICKETS AND I PAY ALL BILLS CONTACT ME FAST FOR DEAL If you want to have deals with good quality please contact me with Contact Info yahoo ID: worldbank.hacking email : worldbank.hacking@yahoo.com ICQ : 645871818 SOFTWARES AT good take away prices CC fullz info, CC DOBDomain hosting. PRICE LIST ALL CVV CVV2. 1 US ( visa,master) =3D 3$/ ( buy > 50 Price $1.2/ 1cvv) 1 US (Amex,dis) =3D 5$/ ( buy > 50 price $3 /1cvv) 1UK =3D 5$/ ( Buy > 50 pric...

Is there a way to suppress DSNs using a mailer flag when a mailer fails? (Especially the PROG mailer.)
I have certain pseudo-mailboxes that use the "PROG" mailer to perform certain functions. However, if these fail, I do not want a DSN sent back to the sender - but I would prefer a syslog entry indicating failure. I looked for a mailer flag to suppress the DSN function (or to ignore the exit code of an external program) and found none. For the moment, I suffix these entries (in aliases) with "||exit 0" which forces the subshell to return an exit code of zero. However, that doesn't give me a syslog entry indicating failure. I also consider this method a bad hack. Am...

formatted I/O #2
I've a simple question about formatted I/O with a recurring data format in each line. E.g. I want to read in ascii data in which each record has a recurring pattern such is parameter, hour, value(hour) for each of the 24 hours in a day. My question is do I have to explicityly provide a format sequence for all 3 fields (parm. hr. val.) repeated 24 times, or is there a way for me to specify 24(parm. hr. val)? Thanks sss In article <ef56844.-1@webcrossing.raydaftYaTP>, sss <stu_schwartz@umbc.edu> wrote: >I've a simple question about formatted I/O with a recurring data ...

I/O calls from Oracle to OS
This post got lost (I can't see it published) so I'm sending it again. Sorry if you'll find it duplicated. Hello, probably this is more a sysadmin question than a DBA one but how can I check if an I/O call from oracle is divided in several reads (or writes) by the operating system? example: if oracle asks at the OS for a 1M block and the maximum I/O blok of my operating system is 512K may I monitor the two different reads? How? I'm most interested in Unix but if you know a common way... Thanks Fabrizio Magni FM <fabrizio.magni@mycontinent.com> wrote in message...

Upgrade PIX515E to PIX 7.0(1) = Flash I/O Error
Hello, I am trying to upgrade a Cisco PIX 515E from 6.3(4) to version 7.0(1). After installing version 7.0(1) binary I reboot the PIX. The system reboots and attempts to write the activation key to flash (which fails with an I/O error), and then it attempts up update the flash filesystem. The flash filesystem update never completes. The error message is "inconsistency in flash found on second pass", followed by "error formatting flash". I have attempted to upgrade several times now using the basic method and the monitor mode. For each iteration I am careful to erase t...

Server monitoring tools (cpu load, mem, disk i/o)
Folks, Looking for some Debian Sarge Linux monitoring tools that track the load (cpu) state of the server (similar to uptime stats), memory (swap, kernel lowfree), disk i/o, and other vital stats that can be viewed remotely via web interface and displayed graphically (trend chart, etc). Network load is good too, but less important since we have some other tools that tell us this (e.g. plesk). Was checking mrtg, rddtool, nagios, cacti, swatch, etc, and concluded these are more about network monitoring out of the box, and less about tracking the server loading state. Also would like to have the...