The intersection of the Firebird and Python communities is small, and I
make next to no attempt to market kinterbasdb. However, I attribute
the slowdown of activity on the kinterbasdb SF forum and the lack of it
here more to the increasing stability of the code and the expansion of
the documentation than to the atrophy of the user community. SF
project stats haven't worked for months, so I can't draw conclusions
from them, but hits to the front page haven't decreased.
I don't have much time to work on kinterbasdb right now anyway, so the
lack of activity suits me fine.
At the top of my kinterbasdb agenda is fixing a bug in the garbage
collection of event queues that can cause a segfault upon process exit.
After that's solved, or if I fail to solve it in a timely manner, I
plan to release 3.1.1, which'll contain a few minor fixes.
For 3.2, here are some planned features listed in descending order by
their urgency to me:
1) Streaming blob support:
2) Better support for concurrency, probably not enabled in default
builds due to the concurrency limitations in the local Windows protocol
implemented by FB <= 1.5.
3) Allow the client programmer the option of explicitly controlling statement
preparation in a manner decoupled from execution. kinterbasdb currently exposes
no way for the client programmer to control the caching of prepared statements;
it's performed behind the scenes, and rather simplistically.
4) More dynamic Unicode support. This one is a low priority for me, so
even after 3.2 development gets underway, it'll probably take another
Jacobi-prompt to trigger a detailed discussion and subsequent
implementation of this particular feature.
A few 3.2 features have already been implemented in CVS HEAD:
- An official implementation of dynamic type translator support for
Python 2.4's standard library decimal module.
- A more convenient alternative to Connection.database_info, called
Connection.db_info, was contributed by Pavel Cisar.
I don't know when I'll get underway with the implementation of 3.2 in
earnest, but it'll probably be some months yet (circa October 2005,
maybe?). I hope to find the time to implement streaming blob support
sooner, since its absence is a serious problem for some users rather
than merely an inconvenience.
Wednesday, May 04, 2005
The intersection of the Firebird and Python
David Rushby wrote on firebird-python group: