FreeTDS Development Group

[Brian Bruns <camber AT ais.org>]

On Mon, 3 Mar 2003, Lowden, James K wrote:

> > From: Brian Bruns [mailto:camber AT ais.org]
> > Sent: March 3, 2003 1:31 PM
> > 
> > DBPROCESS *dbproc = 0;
> > dbinit();
> > dbsetopt(dbproc, ...);
> > 
> > which core dumps because dbsetopt() accesses the null pointer 
> > of 'dbproc'.  
> ...
> > 
> > I have a local patch that 
> > applies the option to all open dbproc's but I suspect that is wrong.
> 
> I suspect it's right.  Microsoft says the same thing Sybase does: "If dbproc
> is NULL, the option is set for all active DBPROCESS structures."  It goes
> on:
> 
> "This function does not immediately set the options specified. They are not
> set until the command buffer is sent to SQL Server (by invoking the
> dbsqlexec function). An additional result for each option set is returned
> through the dbresults function."
> 
> If your patch detects a null dbproc and iterates over the active ones when
> it sees one, I'd say it's right.  As best I can tell, that would conform to
> the documentation.  
> 
> I'd do it recursively in dbsetopt():
> 
>       if dbproc == NULL
>               foreach connection
>                       dbsetopt g_dblib_ctx.connection_list[i] ...
> 

Yeah that's pretty much it.  I'd like to get confirmation on behaviour 
before going ahead with the patch; I'll try to wrap up a test case to run 
on OpenClient before I do.

Brian