Archives 2004

Something in the code introduced to improve the discovery process has managed to produce a failure where the main operations block for too long, causing the zombie watcher to attempt to restart, but it's unable to kill the main process.

Have to figure out a reliable way to force the killing... also need to figure ...

Cinemon was displaying some strange behaviour during discovery, mostly dropping off modems and/or discovering them only after 4 or 5 scans. Haven't been able to isolate the proximate cause yet, but the effect seems to have been fixed.

I hate it when a fix works but I don't know why.

I use the DBAPI's pyformat query format exclusively. It's supported by both the PostgreSQL and MySQL drivers, and it's just a superior mechanism for creating queries (especially with PyTable's SQLQuery class). So, of course, as I was writing the SQLite driver for PyTable I was obviously using pyformat queries.

Problem was, SQLite doesn't support them, ...

SQLite has this annoying habit of reporting all syntax errors with just the text "error preparing statement". That becomes a serious PITA when you're auto-generating huge query strings, but even with the short schema-discovery queries I'm finding it frustrating.

SQLite is a fairly straightforward engine, but it feels far less evolved than PostgreSQL, or even MySQL (which itself feels less evolved than PostgreSQL). It provides a set of "pragma" commands to make it easy to query the current database schema (perform introspection).

The thing about these attempts to make it easier to do ...

Alexei and I used to hang around a lot. What with him moving to China we don't seem to do that any more. He's a renaissance type of person, with a strong focus on the artistic side. He just put up a site showing some of his works. Enjoy it at your leisure, gentle readers.

Didn't wind up having Natasha or Lara come over to grams' place with me, so it was a one-on-one tournament. Of course, grams wound up bragging to all and sundry about how she was going to beat me, so the universe decided to teach her a lesson.

Final score was 4 games to 1 (for ...

How does one go about testing/verifying 183 extensions when one only has 15 or so extensions available on one's graphics card?

Simple, one does not.

I'm going to write a few tests to make sure that the general approach works, but I'm intending to let users who care about particular extensions write tests and submit ...

Still working on getting the OpenGL extensions to compile. The bulk of the extensions compile fine, but there are new pointer-types being defined that are causing problems. I would guess I could just manually copy the appropriate definitions into the swig files, I just want to avoid too much manual maintenance of these files.

183 ...

Finished moving the (successful subset of the) PyOpenGL 2.1 dev branch into the 2.0.x maintenance branch. It's a considerable number of internal changes, hence bumping to 2.0.2 instead of just 2.0.1.10.

Biggest change is using the new SWIG, which causes ripple effects throughout the codebase. Not a lot of user-visible changes, really, but oh well. ...