Re: [Gnumed-devel] getting closer to a version that "makes sense"

[Tim Churches]

Karsten Hilbert wrote:
> On Sat, Apr 09, 2005 at 06:08:27AM +1000, Tim Churches wrote:
> 
> 
>>Last time we looked, psycopg didn't have any tests at all, which is why
>>we choose pyPgSQL instead.
> 
> Do tell !?!

Well, that was nearly 2 years ago, and psycopg may have acquired some
tests since then. But I now well and truly believe in the absolute
necessity of unit tests which are written at the same time as the actual
code - repeatedly code which we have been certain is correct is shown to
fail in a "corner case" unit test. It's *very* desirable that different
people write the code and the tests, though.

>>I suspect that the baby is in danger of being thrown out with the
>>bathwater.
> 
> There ain't no danger yet :-)  It's not like we are going to
> actually *do* anything today or tomorrow ...
> 
>>The problem with pyPgSQL is that the developers aren't very
>>responsive to emails and seem reluctant to include patches they didn't
>>write, and they have not rolled up a tarball or made a Windows binary
>>installer for nearly two years (eg there is no Windows installer for
>>pyPgSQL for Python 2.4 available). However, the code is being maintained
>>in CVS - there are commits quite regularly.
> 
> I see. Good to know.
> 
> 
>>So its really just a matter of taking the code from CVS and creating
>>your own tarball and/or binary installer. That's what we plan to do for
>>NetEpi in the next two months.
> 
> Let's "pool resources" in this case. Maybe I am dumb but it
> sounds more manageable than rolling our own entirely.

We'll be starting either one or two contractors in the next two weeks to
do various Python work, and if there is time, I'll get one of them to
look at making fresh installation packages (and maybe a Windows binary
installer) from the latest CVS code, as well as doing more testing
against PG 8.0

The Mozilla/GPL license incompatibilities are not an issue with pyPgSQL,
since it uses teh Python license, so no impedimentto close co-operation
on this.

Tim C