Recommendations for a PostgreSQL db adapter, please?

[Gerhard Häring]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Tim Churches wrote:
> Gerhard Häring wrote:
>>[...] Considering pyPgSQL, psycopg1, PyGreSQL and psycopg2 - psycopg2 is a good
>>choice.
> 
> Hmmm, Gerhard, you are listed as one of two developers for pyPgSQL on
> SourceForge  ( http://sourceforge.net/projects/pypgsql/ ). The fact that
> you are recommending a different DB adaptor project suggests a degree of
> disengagement with pyPgSQL.

It's not that I'd now say that pyPgSQL is bad software, but as you say it
didn't see a lot of maintenance the last two years, and AFAIK this won't
change anytime soon.

With the last release, it was pretty much feature complete, and Billy G.
Allie and me were talking about a rewrite of many parts in C for increased
performance, and for taking advantage of the new PostgreSQL API. I told him
that I was unfortunately too busy with other things to contribute
significantly to pyPgSQL. I see myself more as a helper for pyPgSQL nowadays.

> Is Billy G. Allie still interested in pyPgSQL, or is it now completely
> unloved?

AFAIK he is still interested and committing fixes to the current codebase
now and then and working on the rewrite.

I don't know, however, if there's a plan for the long overdue maintenance
release to the last release from 2003.

> If so, then perhaps some new maintainers can be found for it [...]

I'll just cc Billy so he can comment :-)

> - several projects rely on pyPgSQL,
> including ours ( http://sourceforge.net/projects/netepi/ )  - for which
> we currently need to provide our own pyPgSQL tarball rolled from CVS
> plus some of our own minor but important patches - which were submitted
> for consideration but have not (AFAIK) been checked into the pyPgSQL
> CVS. The latest tarball for pyPgSQl available from the pyPgSQL
> SourceForge pages is dated 2003.
> 
> No criticism is intended in any of the foregoing observations - it is
> inevitable that people move on to new projects (such as pySQLite), but
> it would be a shame if pyPgSQL just rotted, because it has several
> things in its favour, such as some unit tests (which were conspicuously
> absent from any of the alternatives when we evaluated them in 2003 -
> perhaps they have been added by now). Overall we have found pyPgSQL to
> be very reliable.

- -- Gerhard
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFEU3l/dIO4ozGCH14RAm7uAJ9AE5TnPvDL6Ic0z6Cf+QSSlkWnRwCgirdy
TQnyqOyWMGX2KbGE+gFKIZU=
=cTbz
-----END PGP SIGNATURE-----