FreeTDS Development Group

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

On Wed, 31 Mar 2004 <michaels AT crye-leike.com> wrote:
> I'd like to
> thank all of the developers and contributors who have made this great
> piece of software available to the community.

        :-)

> I've compiled the current stable version of FreeTDS (0.62.1) [1], along
> with PHP 4.3.3 using the --with-sybase option [2].  I'm noticing a
> problem with the return value of PHP's mssql_query() function when the
> query calls a stored procedure that returns an empty result set (0
> rows).  Normally the return value of this function should be a result
> resource, but in this case the function is returning a boolean true.  

Based on what you found, and on my knowledge of our db-lib implementation,
it's very hard to see how (or if) this is caused by FreeTDS.  The
--with-mssql extension resolves to the very same db-lib call, and works
correctly.  Perhaps the older with-sybase handling expects dbresults() to
behave a little differently?  

I hope someone here who's more familiar with PHP can help you.  

> I tried the latest snapshot as of yesterday (0.63.dev.20040329), just to
> see if the issue was addressed there, but unfortunately I cannot get PHP
> to compile against it (due to various "undeclared" errors with
> SYBDATETIMN, SYBFLTN, SYBINTN, and SYBMONEYN).  

That's looks like a problem.  We're pruning our header files; maybe we've
gone too far.  

> So, it
> appears that this problem was introduced somewhere between versions
> 0.61.2 and 0.62.1.

You will find 0.62.2 in our current directory (next to the snapshot).  I
haven't officially announced it yet, but as no one's complained, I guess
it will become official soon.  

> At the time I first starting using PHP with FreeTDS, it was generally
> recommended to go with --with-sybase and dblib, as this was considered
> more mature than ctlib.  For kicks, I decided to compile PHP
> using --with-sybase-ct against 0.62.1, and the problem does not appear
> in that configuration.  So it seems that using sybase-ct is now an
> option. Does anyone know if this combination (PHP/FreeTDS/ctlib) is now
> considered mature enough for production servers?  This isn't strictly a
> FreeTDS question, because my understanding was that PHP's ctlib
> extension was also a bit rough a while back...

I don't have any opinion about PHP, because I don't use it.  I can say
that our ct-lib is much further along, almost unrecognizably changed,
since 0.53.  It's used by DBD::Sybase (Perl) with pretty good success.  

> Also, thanks to the list I recently discovered that it's possible to
> compile PHP using --with-mssql against FreeTDS.  I tried this and the
> problem also does not appear.  It's interesting to me that this approach
> seems to also use dblib, at least according to ldd, but the error does
> not occur.  This suggests to me that this may not be a FreeTDS bug, but
> may be an issue with the way PHP's sybase extension interacts with
> FreeTDS, since the mssql extension seems to work.

I agree.  I'm not saying it's a PHP bug; I wouldn't be surprised if it
turned out to be some way in which what we're doing is slightly different
than what Sybase's db-lib does.   Someone will have to look at the PHP
code to see why it thinks it should return a boolean when there are no
results.  

> So, does anyone have an recommendations on what I should do?  

I'm going to leave that to the PHP experts here, except to tell you that
many people successfully use the msssql extension.  

> I can also provide additional information, including
> sample scripts, if required.  

If you want to add a sample script to our samples/ directory, or extend
what's there, I'll be happy to include it.  Eventually, we really should
have a PHP regression test.  

> [1] Configured
> with: --with-tdsver=4.2 --enable-msdblib --enable-dbmfix --with-gnu-ld
> --ena ble-shared --enable-static

--enable-dbmfix has no effect anymore.  

Nice to hear from a long time happy user.  Too bad you had to write.  ;-)

--jkl