CDX indexes - comp.lang.clipper

Hi, Having spent many years using just NTX indexes. I was starting to have speed issues with a busy client so I decided to change to CDX indexes. I have used the CDX indexes only on the major files that have multiple indexes and the speen incease is significant. This all went well apart from some bizarre stuff, for example: I had 6 records indexed on a numeric field, 1000.00 through to 1000.05 X := 1000 dbseek(x) // found x += 0.01 dbseek(x) // found x += 0.01 dbseek(x) // found x += 0.01 dbseek(x) // found x += 0.01 dbseek(x) // NOT found but why? dbseek(1000.04) // found x += 0.01 dbseek(x) // NOT found but why? dbseek(1000.05) //found This had been working for many years with the NTX index! I ended up indexing on STR(LABNO,9,2) and changing all the seeks and it all worked OK, but why? My main question is regarding the DBFCDX and SIXCDX drivers. Are CDX indexes created by both systems interchangable? I ended up using the SIXCDX driver as it seemed to work better (too complicated to explain). I also need to use xBase++ for some of the serial comms stuff that I do as Clipper Tools III does not recognise non-standard (USB and add in card) serial ports. How do I use both CDX and NTX indexes in xBase++ programmes? What combination of DBELOAD and DBEBUILD to I need to acheive this? Many Thanks Ian Boys DTE Systems Ltd

Ian, This is all to do with decimal rounding and precision. Doing a seek on a float is always going to be difficult because internally in the DBF (or in the variable), the actual value could be 1000.0400000000001 or 1000.03999999999999 etc. Make sure of your default settings for SET DECIMAL and related items. Geoff "Ian Boys" <not@given.com> wrote in message news:bJKdnf6ufJYBcPbWnZ2dnUVZ8kWdnZ2d@bt.com: > Hi, > Having spent many years using just NTX indexes. I was starting to have speed > issues with a busy client so I decided to change to CDX indexes. I have used > the CDX indexes only on the major files that have multiple indexes and the > speen incease is significant. This all went well apart from some bizarre > stuff, for example: > I had 6 records indexed on a numeric field, 1000.00 through to 1000.05 > X := 1000 > dbseek(x) // found > x += 0.01 > dbseek(x) // found > x += 0.01 > dbseek(x) // found > x += 0.01 > dbseek(x) // found > x += 0.01 > dbseek(x) // NOT found but why? > dbseek(1000.04) // found > x += 0.01 > dbseek(x) // NOT found but why? > dbseek(1000.05) //found > This had been working for many years with the NTX index! I ended up indexing > on STR(LABNO,9,2) and changing all the seeks and it all worked OK, but why? > > My main question is regarding the DBFCDX and SIXCDX drivers. Are CDX indexes > created by both systems interchangable? I ended up using the SIXCDX driver > as it seemed to work better (too complicated to explain). > > I also need to use xBase++ for some of the serial comms stuff that I do as > Clipper Tools III does not recognise non-standard (USB and add in card) > serial ports. How do I use both CDX and NTX indexes in xBase++ programmes? > What combination of DBELOAD and DBEBUILD to I need to acheive this? > > Many Thanks > > Ian Boys > DTE Systems Ltd

i always use str() indexing numerfields index on str(field,16,2) dbseek( str( nseek ,16,2) ) Ian Boys schrieb: > Hi, > Having spent many years using just NTX indexes. I was starting to have > speed issues with a busy client so I decided to change to CDX indexes. I > have used the CDX indexes only on the major files that have multiple > indexes and the speen incease is significant. This all went well apart > from some bizarre stuff, for example: > I had 6 records indexed on a numeric field, 1000.00 through to 1000.05 > X := 1000 > dbseek(x) // found > x += 0.01 > dbseek(x) // found > x += 0.01 > dbseek(x) // found > x += 0.01 > dbseek(x) // found > x += 0.01 > dbseek(x) // NOT found but why? > dbseek(1000.04) // found > x += 0.01 > dbseek(x) // NOT found but why? > dbseek(1000.05) //found > This had been working for many years with the NTX index! I ended up > indexing on STR(LABNO,9,2) and changing all the seeks and it all worked > OK, but why? > > My main question is regarding the DBFCDX and SIXCDX drivers. Are CDX > indexes created by both systems interchangable? I ended up using the > SIXCDX driver as it seemed to work better (too complicated to explain). > > I also need to use xBase++ for some of the serial comms stuff that I do > as Clipper Tools III does not recognise non-standard (USB and add in > card) serial ports. How do I use both CDX and NTX indexes in xBase++ > programmes? What combination of DBELOAD and DBEBUILD to I need to > acheive this? > > Many Thanks > > Ian Boys > DTE Systems Ltd > >

Geoff, I had kind of figured that but I thought that the set decimal didn't actually change the value, just how it was displayed and therefore would not help the situation. It's still strange that it had been working for years and only came to light with the index change. Ian "Geoff Schaller" <geoffx@softxwareobjectives.com.au> wrote in message news:cBSan.5741$pv.2039@news-server.bigpond.net.au... > Ian, > > This is all to do with decimal rounding and precision. Doing a seek on a > float is always going to be difficult because internally in the DBF (or in > the variable), the actual value could be 1000.0400000000001 or > 1000.03999999999999 etc. Make sure of your default settings for SET > DECIMAL and related items. > > Geoff > > > > "Ian Boys" <not@given.com> wrote in message > news:bJKdnf6ufJYBcPbWnZ2dnUVZ8kWdnZ2d@bt.com: > >> Hi, >> Having spent many years using just NTX indexes. I was starting to have >> speed >> issues with a busy client so I decided to change to CDX indexes. I have >> used >> the CDX indexes only on the major files that have multiple indexes and >> the >> speen incease is significant. This all went well apart from some bizarre >> stuff, for example: >> I had 6 records indexed on a numeric field, 1000.00 through to 1000.05 >> X := 1000 >> dbseek(x) // found >> x += 0.01 >> dbseek(x) // found >> x += 0.01 >> dbseek(x) // found >> x += 0.01 >> dbseek(x) // found >> x += 0.01 >> dbseek(x) // NOT found but why? >> dbseek(1000.04) // found >> x += 0.01 >> dbseek(x) // NOT found but why? >> dbseek(1000.05) //found >> This had been working for many years with the NTX index! I ended up >> indexing >> on STR(LABNO,9,2) and changing all the seeks and it all worked OK, but >> why? >> >> My main question is regarding the DBFCDX and SIXCDX drivers. Are CDX >> indexes >> created by both systems interchangable? I ended up using the SIXCDX >> driver >> as it seemed to work better (too complicated to explain). >> >> I also need to use xBase++ for some of the serial comms stuff that I do >> as >> Clipper Tools III does not recognise non-standard (USB and add in card) >> serial ports. How do I use both CDX and NTX indexes in xBase++ >> programmes? >> What combination of DBELOAD and DBEBUILD to I need to acheive this? >> >> Many Thanks >> >> Ian Boys >> DTE Systems Ltd >

Dear Ian Boys: On Feb 5, 4:14=A0am, "Ian Boys" <n...@given.com> wrote: > I had kind of figured that but I thought > that the set decimal didn't actually change > the value, just how it was displayed and > therefore would not help the situation. >=A0It's still strange that it had been working > for years and only came to light with the > index change. Not really that odd, since the index RDD could have different logic for data handling. And since it was known to be a problem... David A. Smth

Hello, as far as I remember you must use STR() on numbers if you use CDX. It assumes fixed sizes of the indexexpression, maybe because of multiple tags in one file. HTH Tom "Ian Boys" <not@given.com> schrieb im Newsbeitrag news:bJKdnf6ufJYBcPbWnZ2dnUVZ8kWdnZ2d@bt.com... > Hi, > Having spent many years using just NTX indexes. I was starting to have > speed issues with a busy client so I decided to change to CDX indexes. I > have used the CDX indexes only on the major files that have multiple > indexes and the speen incease is significant. This all went well apart > from some bizarre stuff, for example: > I had 6 records indexed on a numeric field, 1000.00 through to 1000.05 > X := 1000 > dbseek(x) // found > x += 0.01 > dbseek(x) // found > x += 0.01 > dbseek(x) // found > x += 0.01 > dbseek(x) // found > x += 0.01 > dbseek(x) // NOT found but why? > dbseek(1000.04) // found > x += 0.01 > dbseek(x) // NOT found but why? > dbseek(1000.05) //found > This had been working for many years with the NTX index! I ended up > indexing on STR(LABNO,9,2) and changing all the seeks and it all worked > OK, but why? > > My main question is regarding the DBFCDX and SIXCDX drivers. Are CDX > indexes created by both systems interchangable? I ended up using the > SIXCDX driver as it seemed to work better (too complicated to explain). > > I also need to use xBase++ for some of the serial comms stuff that I do as > Clipper Tools III does not recognise non-standard (USB and add in card) > serial ports. How do I use both CDX and NTX indexes in xBase++ programmes? > What combination of DBELOAD and DBEBUILD to I need to acheive this? > > Many Thanks > > Ian Boys > DTE Systems Ltd > >

hi, > I also need to use xBase++ > How do I use both CDX and NTX indexes in xBase++ programmes? What > combination of DBELOAD and DBEBUILD to I need to acheive this? please ask in Alaska Newsgroup again while Cl*pper have RDD other than Xbase++ DATA and ORDER Component. greetings by OHR Jimmy p.s. Xbase++ Version, OS(), Network, Ops lock

Similiar Articles:

file opening slows down - puzzling - comp.lang.xharbour
> > > CDX indexes are faster and more reliable, mainly with regard to memo > > fields. I think not many people here still use .NTX. Nobody perhaps.

Problem setting index (Bug?) - comp.lang.xharbour
... and SIX3, but I wrote a quick testbed to recreate the index in xHarbour, using the code: INDEX ON DBF_NAME+STR(FIELD_NUM) TAG "DBF" TO "..\xHarb\Index\DATADICT.CDX ...

Dbseek problem - comp.lang.xharbour
All the fields I use in the index are critical for the compare. I thought I read somewhere that the key length for cdx was 220. If I am wrong, what is the key ...

Credit default swap index - Wikipedia, the free encyclopedia
CDX.NA.HY.BB: 37: Index of high yield CDSs with a BB rating: CDX.NA.HY.B: 46: Index of high yield CDSs with a B rating: CDX.NA.XO: 35: CDSs that are at the crossover point between ...

Dow Jones CDX Indexes Definition | Investopedia
Dow Jones CDX Indexes - Definition of Dow Jones CDX Indexes on Investopedia - A series of indices that track North American and emerging market credit derivative indexes.

7/17/2012 8:48:21 AM