FreeTDS Development Group

["James K. Lowden" <jklowden AT freetds.org>]

Rich Young wrote:
> is it possible that we're missing something more obvious
> than all this?  I mean, if I have a query that works in tsql and isql,
> but fails when it's part of a PHP script, does that prove it's a PHP
> bug?  

That it runs in tsql proves next to nothing, that FreeTDS is internally
consistent and can communicate with the server, which nowadays we expect. 


That it runs is isql proves something: isql is using FreeTDS as the ODBC
driver, and works correctly.  So, yes, you have one ODBC client that is
satisfied with FreeTDS's behavior.  That's to the good.  If I were in your
shoes, I'd take my query and my isql results (and my stand-alone PHP
script failure) to the PHP folks and ask them what to do.  

The possibility remains that the problem lies on the boundary of unixODBC
and PHP, possibly tickled by something in FreeTDS.  It's impossible to
know from here.  I hesitate to call it a bug without knowing what the
documentation says and which program (if any) is not conforming to its
documented behavior.  

But, yes, I'd say you've exhausted your search here.  From what I see, the
problem manifests itself in the upper layers, and it's certainly PHP that
segfaults.  Best would be if they can help you track down the wild
pointer.  That would at least lead you in the right direction.  

HTH.  

--jkl