On Thu, Oct 13, 2005 at 07:35:53AM -0400, Moon, John wrote:
> From: Tim Bunce [mailto:[EMAIL PROTECTED]
> Sent: Friday, October 07, 2005 12:19 PM
> To: Jared Still
> Cc: Moon, John; 'dbi-users@perl.org'
> Subject: Re: Execute failed: ORA-01026: multiple buffers of size > 4000 in
> the bind list (DBD: oexec error)
>
> On Fri, Oct 07, 2005 at 07:02:43AM -0700, Jared Still wrote:
> > On Wed, 2005-10-05 at 10:46, Moon, John wrote:
> > > The following code produces this error - ORA-01026 - after 2,000
> inserts...
> > > Any help or suggestions will be greatly appreciated.
> >
> > You should troubleshoot this with the help of Oracle Support if you have
> > it.
> >
> > There are a number of possible causes, the two most frequent being
> > incorrect NLS_LANG environment variable settings, or bugs.
>
> Though I suspect there may be a DBD::Oracle bug where (for example) a
> buffer is growing by one byte each execute until it reaches the 4000
> limit and causes the error.
>
> This error has been reported on and off for years. No one has produced a
> small enough test case to make it easy to track down and see if the
> problem lies with DBD::Oracle or not.
>
> Tim.
>
> jwm wrote.
>
> Thanks to all who replied...
>
> The program I have failed consistently at around 2,000 inserts... In trying
> to resolve the problem (thinking it may have been a data error in one record
> (?)) I put a trap in to start printing (dump) the records at the point of
> failure... The program did not fail. I took the print out and program would
> fail.
Interesting.
> I have implemented the following work around and have not had a failure
> since...
> &MyStd_Titles("Loaded $item_nbr",35) if $item_nbr%100 == 0;
> $item_nbr++;
>
> (the sub called prints a timestamped message...)
>
> Interestingly Sql Loader from Oracle does a similar message...
>
> If anyone with more knowledge that I (which is very little) would like to
> pursue this problem and give me some directions on how/what to do to product
> results for someone to look at I will be happy to try at this site.
It would be a *big* help if you could distill the code down to a very
small test case. The smaller/simpler the test case, the faster the fix :)
Tim.
