Обсуждение: ODBC - Makefiles

Hi,

I am confused. I have the zip file distribution from Byrons web site.
Its supposed to be the latest and greatest but it has no make files for
UNIX/Linux?

I also have a set of source which is integrated into the server
distribution. It appears to have UNIX/Linux make files but I could spend
way too much time getting my head into it then I want to. I have many
other drivers to work on and I can not spend time figuring out every
source/config structure of the servers etc.

My preferred option is to work with the zipped/standalone version and
then hand the whole thing back to Byron... things brings me back to
'where are the UNIX/Linux make files'?

Peter Harvey

NOTE: I can code up a storm but I am NOT a config/install guru!

> I am confused. I have the zip file distribution from Byrons web site.
> Its supposed to be the latest and greatest but it has no make files for
> UNIX/Linux?

Right. Byron develops under WIN32. Several of us helped port it to
Unix some time ago.

> I also have a set of source which is integrated into the server
> distribution. It appears to have UNIX/Linux make files but I could spend
> way too much time getting my head into it then I want to. I have many
> other drivers to work on and I can not spend time figuring out every
> source/config structure of the servers etc.

?? There is one directory which contains an older version of Byron's
code. He is planning on updating our CVS repository, though it looks
like he hasn't had a chance to do that yet. The Right Way (tm) to
build a standalone version of the driver on Unix is to build a
standalone tar file from a configured Postgres tree:

cd src
./configure --with-odbc
cd interfaces/odbc
make standalone

Take the resulting tar file and put it into a separate area, then make
from there.

> My preferred option is to work with the zipped/standalone version and
> then hand the whole thing back to Byron... things brings me back to
> 'where are the UNIX/Linux make files'?

I'd suggest starting from an updated standalone tar distribution as
described above. I'll build something and send it to you when Byron
has a chance to do the updates (and I can test them), or you can take
the existing standalone distribution on ftp.postgresql.org and merge
Byron's code in yourself. He had suggested waiting for his changes
however.

btw, since the newest code is *not* in the cvs repository, folks doing
development under Unix can not test or keep up to date with changes.
So we should probably consider the current cvs tree as the best
available distribution, even though it is out of sync with zipballs
posted at insightdist.com. Is there anyone out there who is actively
developing ODBC under Unix and perhaps WIN32 who would want to help us
keep up to date with changes for the Unix side of things? It's usually
not much work, but it could use someone with that focus...
                      - Tom

-- 
Thomas Lockhart                lockhart@alumni.caltech.edu
South Pasadena, California

> Ok. I am going to merge them and then update to work with unixODBC. I guess I
> will have to fire the source off to you at that time?

That will probably work. I'll forward it to Byron and Cary after
trying a build here. Obviously we will want to ensure compatibility
and maintainability with our existing supported interfaces and
environments...

Byron, we should also have your existing code checked in and tested
under Unix in preparation for the v6.5 release. That is probably more
important, timing wise, than incorporating any new patches from Peter
since his effort is still in an alpha/beta stage. If it is available
and tested soon, we could ask Peter to remerge his work (should be
trivial with diff and separate source trees) to ensure a consistant
code set.
                          - Thomas

-- 
Thomas Lockhart                lockhart@alumni.caltech.edu
South Pasadena, California