Archives week 36 of 2006

Spent the full work-day today on OpenGL-ctypes. The glGet* variables are now getting automatically registered. Image types, it turns out, seem to already be registered. I've also started work on making the docstrings more reasonable, little things like including the extension's Overview section in the docstring (so you can tell what the extension is intended ...

We installed Ubuntu on Soni's machine a while ago, forgoing Gentoo because we figured she should start with a "consumer" oriented distribution. Problem is, first time she runs into a problem I haven't the confidence to fix it by "remote control" (as it were).

In particular, Eric3, though it shows up on the Ubuntu web-site ...

Spent the rest of the day on the auto-generation system for producing GL extension wrappers. Mostly I'm looking at parsing the extension definition files and turning them into instructions to do such things as registering glGet* extensions, automatically registering image types for support, that kind of thing. Also tweaked the code so that marking a ...

This morning I decided to eliminate the _strings.so C module from OpenGL-ctypes. This module had a single function which retrieved the data-pointer from a Python string structure. That was done mostly because of things like potentially having a different Python version or a different Python compilation flag resize the PyObject structure header and wanting to ...

Spent the whole day again on sysadmin work. Managed to get the VoIP sub-application installed and functioning, but I still don't have a fully functional ISP Billing System.

Discovered that the new (long promised) schema update for the ISP Billing System is going to cause dozens of (luckily fairly trivial) changes throughout the VoIP sub-application, ...

Spent the whole day setting up our new development server (save for about 1/2 hour on billable work). I've got all of the secondary services set up, I'm just working on getting the legacy billing system built and set up now. Lot of dependencies in the tree, including multiple local-only data-types, a weird macro-expansion tool ...

The bug in the array handling for OpenGL-ctypes was due to my trying to be too clever. Oops. Finished running the OpenGL and OpenGLContext test suites. Biggest issues that surfaced were that the nurbs support has been broken somehow :( , and that the OpenGLContext mouse-handling code is accepting far too many events for processing, ...

Running the PyOpenGL GLUT demo "gears.py", which is basically identical to the standard glxgears program, today I get 1430fps on my workstation, (compared to about 1600 for glxgears today). This same machine was getting around 150fps last week (with either program). I'm guessing the complete system rebuild with gcc 4.x and glibc 2.4.x (which was ...

Spent the lunch break working on testing OpenGL-ctypes. Got through all of the PyOpenGL demos, but far more were failing than should have been. It appears I've introduced a bug in array handling at some point and didn't notice it. Grr. Worse, as I was testing in the OpenGLContext area I discovered a bug that ...

Decided that I just had to get some work done on OpenGL-ctypes if it's going to come out any time reasonably soon after Python 2.5 final. Accordingly took the day off and worked solely on OpenGL-ctypes. Primary task today was to work on the documentation for developers and contributors. I've posted that (in-process) documentation to ...