index corruption dos/clipper harbour/linux - Clipper

This is a discussion on index corruption dos/clipper harbour/linux - Clipper ; I use several databases (standard dbf/cdx) whom are both used by a clipper (5.3) application in a windows environment as well as an harbour(45) on an linux slackware environment. The databases are situated on the linux machine, and for the ...

+ Reply to Thread
Results 1 to 4 of 4

index corruption dos/clipper harbour/linux

  1. Default index corruption dos/clipper harbour/linux

    I use several databases (standard dbf/cdx) whom are both used by a clipper
    (5.3) application in a windows environment as well as an harbour(45) on an
    linux slackware environment. The databases are situated on the linux
    machine, and for the clipper app. accessible via samba.

    Frequently, (but not always) i encounter 'corruption detected' errors. When
    only using the clipper-app, it works fine (in many multi-user sessions).
    The same applies for the harbour app.

    In the harbour app, after every UNLOCK id do
    a COMMIT and a __Run("sync;sync;sync;")

    Is there a different locking scheme? Am i missing something?

    The clipper app. is very large and it will take a very long time for me to
    re-code it al to harbour. I use the harbour app. for telnet sessions to Rf-
    Barcode terminals.

    Any clues?
    Emmer

  2. Default Re: index corruption dos/clipper harbour/linux

    Hi Emmer,

    > I use several databases (standard dbf/cdx) whom are both used by a clipper
    > (5.3) application in a windows environment as well as an harbour(45) on an
    > linux slackware environment. The databases are situated on the linux
    > machine, and for the clipper app. accessible via samba.


    I don't really have a good suggestion to give. However, this is
    something I intend to do in the near future, so it's quite important
    for me to follow this issue.

    >
    > Frequently, (but not always) i encounter 'corruption detected' errors. When
    > only using the clipper-app, it works fine (in many multi-user sessions).
    > The same applies for the harbour app.
    >
    > In the harbour app, after every UNLOCK id do
    > a COMMIT and a __Run("sync;sync;sync;")


    Why do you call sync after every commit ? Shouldn't the file system be
    left alone working normally ?

    >
    > Is there a different locking scheme? Am i missing something?

    [snip]

    I suppose you tested that record locking is honoured, either requested
    by clipper or harbour.
    >From my experience, on local (win Me) dbf/ntx, a rlock() request made

    by harbour is not seen by clipper. The reverse is also true.
    Locking appears to work good only when the dbf/ntx is on samba server,
    and a SET DBFLOCKSCHEME = 2 is issued at the beginning of the harbour
    program . But I don't have an history of continuous use, only sporadic
    tests...
    Sorry for not beeing able to help,
    Regards,
    António Vila-Chã
    Viana do Castelo
    Portugal


  3. Default Re: index corruption dos/clipper harbour/linux

    Thanks Antonio,

    Your remark concering record-lock testing was crucial. I must confess that
    I did not actually test record locking being honoured by clipper and
    harbour simultaniously.

    When I did do a record-lock test, it failed. A record lock in Clipper
    DBF/CDX was not honoured by Harbour and vice-versa.

    So, I tried "SET DBFLOCKSCHEME=2" but this was a not valid command in
    Harbour45. (It is supposed to be valid in xHarbour, but I haven't installed
    xHarbour on my system.)

    In my DOS/Clipper application I did not link in the CDXLOCK.OBJ. When doing
    so, it solved all my problems. Record locking now is honoured by both
    dos/clipper & harbour/linux.

    Kind regards
    Emmer...

  4. Default Re: index corruption dos/clipper harbour/linux

    Hi Emmer,

    [snip]
    > So, I tried "SET DBFLOCKSCHEME=2" but this was a not valid command in
    > Harbour45. (It is supposed to be valid in xHarbour, but I haven't installed
    > xHarbour on my system.)

    [snip]

    Sorry, it should be "SET DBFLOCKSCHEME TO 2".

    Regards,
    António Vila-Chã
    Viana do Castelo
    Portugal


+ Reply to Thread

Similar Threads

  1. index corruption dbfntx
    By Application Development in forum xharbour
    Replies: 5
    Last Post: 09-22-2007, 05:00 PM
  2. harbour and ADS from clipper
    By Application Development in forum Clipper
    Replies: 2
    Last Post: 09-01-2007, 09:37 PM
  3. a new seb site regarding clipper/harbour
    By Application Development in forum xharbour
    Replies: 0
    Last Post: 05-28-2007, 04:48 AM
  4. Harbour clipper beta2 Linux Ubuntu quick setup
    By Application Development in forum Clipper
    Replies: 0
    Last Post: 04-07-2007, 03:12 PM
  5. Index corruption
    By Application Development in forum Clipper
    Replies: 6
    Last Post: 10-25-2005, 11:14 PM