Re: parameters missing in RC 64 Bits

On Sun, Apr 20, 2008 at 02:24:18PM -0400, James K. Lowden wrote:
> Federico Alves wrote:
> > I have a perl script that I use to test my application, and executes a
> > stored procedure against SQL 2005, with "OUTPUT" parameters. It works
> > fine in 32 bits but fails to return any parameters in 64 Bits. 
> 
> Igor Chudov wrote:
> > 
> > col_types array ref, which used to work and contained valid type
> > numbers, no longer works on 64 bit ubuntu Hardy. Instead of valid
> > types, it contains a bunch of zeroes (as many as columns).
> 
> Gentlemen, 
> 
> Your questions are too high-level for me to answer without a great deal of
> work.  I don't have a 64-bit environment to test on, and don't even know
> what function, if any, is alleged to fail.  

Whatever I can do to help you, I will.

I actually have some understanding of the issue. 

Ubuntu Hardy uses a different odbc.h where some types have been
redefined for 64 bit.

I had to make some changes to make our odbc C++ code (no relation to
the particular problem I came here with) compile under 64 bit ubuntu.

That c++ code uses freetds-devel package.

These odbc.h changes come from Debian.

Anyway. I can give you a user account on a 64 bit Ubuntu Hardy
computer. (no root). I also do not have a MSSQL server, so you would
need to use your own over the internet. But I will be delighted to
give you an account and install whatever packages you want.

> Do all src/ctlib/unittests work?  Many of them rely on column metadata,
> and rpc_ct_param.c tests output parameters.  If they pass, we at least can
> say ct-lib works as expected.  Not right, necessasrily, but as expected.  
> 
> Problems like this are always a matter of the function behaving as the
> caller expects.  You need to narrow down who the caller is and what it
> expects (and whether that's the right expectation).  Then we can see if
> how the function behaves, and what needs to be changed.  

I will give this a try, what do I need to download and compile?

i