FreeTDS Development Group

["Anwar, Mohammed" <mohammed.anwar AT intel.com>]

Thanks for the replies.

I checked out the latest CVS and the fix is there.

I had the release version which didn't have the updated code.

My code is virtually the same as the example in the link in my message, this 
can be used as a test.

http://manuals.sybase.com/onlinebooks/group-cn/cng1250e/dblib/@Generic__BookTextView/47490;hf=0#X

Thanks again.
Anwar


-----Original Message-----
From: James K. Lowden [mailto:jklowden AT schemamania.org]
Sent: Wednesday, January 29, 2003 11:31 PM
To: freetds AT lists.ibiblio.org
Subject: Re: [freetds] possible bug: bcp_bind seems to be broken when
using variables assignment


On Wed, 29 Jan 2003 22:12:55 -0800, "Anwar, Mohammed"
<mohammed.anwar AT intel.com> wrote:
> Hi,
>       When using bcp_init followed by bcp_bind the bcp_bind will always
>       fail (for me anyway, and I have tried a number of variations on
>       the parameters to bcp_bind).
>       
>       Once I looked at the code I realized why the bind was failing.
> 
>       I looked at the source in bcp.c and it seems bcp_bind has a check
>       for :
>               if (table_column > dbproc->host_colcount)
>                       return FAIL;
>       
> 
>       which will always be return FAIL as the dbproc->host_colcount
>       variable is set to 0 in the bcp_init when calling :
>               _bcp_clear_storage(dbproc);
> 
>       it seems this variable is only updated when the input file is
>       provided and a call is made to 
>               bcp_columns
> 
>       the bcp_init function only updates the dbproc->bcp_colcount
>       variable.

And my bcp_bind documentation says:

        "It is an error to call bcp_columns when using bcp_bind."

Funny, there's no unit test for bcp_bind.  Hmph. 

Mohammed, a quick check of the CVS code confirms what Bill says, that
bcp_init() sets things up correctly.  Rather than a log, could you just
send a little test case that demonstrates the problem?  I'd like to add it
to our unit tests once the problem is solved, one way or the other.  

Regards, 

--jkl