Today was spent working on web content for the DS site. My previous addition of automatically generated date indexes turns out to not actually work. I'm trying to find a way to make it work, without damaging things.

Spent a few hours writing a Python script to convert OpenSim OAR files to POVRay scenes. I consider this a learning exercise, although it could turn out to be a fun way of externally viewing a region. I currently have terrain R32 to POVRay mesh, although the triangulation isn't the same as in OpenSim. And it's uncolored. There is also blue transparent water. I also have sphere and cube prims being positioned and scaled. I don't have coloring for them yet either. I haven't tried texturing yet, although the liklihood of POVRay supporting JPEG2000 format textures isn't looking too likely.

I find the idea of a POVRay based viewer for OpenSim to be amusing. It's all primitive shapes, after all...

Continued working on the EvoGrid, and have reached a significant milestone that implements the core of the "ChemoGrid" design, separating simulation and analysis with a web service coordinating the two.

Spent some time setting up some "new" (gifted by a friend when replaced) hardware to act as an EvoGrid simulator box. The BIOS on the machine decided it didn't like my RTL8139D based NIC (this may be related to plugging it into a 64bit PCI slot first, the card may have been damaged). The kernel would report (via dmesg) that the device was being ignored, because of "unknown header type 10". This means that the card is reporting itself as a "reserved for future use" class of device. No changes I was able to make in the BIOS caused this card to detect (such as adjusting PCI timing), but I was able to replace it with another. The machine is now a Debian 5.03 base install, with OpenSSH installed, and will be used to port the EvoGrid codebase from the currently supported Debian version (4) to current, as well as running simulation time.

Work has been full on EvoGrid development. See the evogrid-dev mailing list for more details.

Recently moved the wireless network to WPA2 and remembered why I'd tried it before and reverted to WEP. My laptop's ndiswrapper'ed wifi add-on card doesn't reliably connect to WPA2 networks. It keeps dropping into scanning mode after associating. Except when it does work.

Fell out of the habit of posting with December's social whirl.

Investigated a C# port of libnoise to test and confirm it produces the same noise. Some of the functionality was missing and noiseutils had not been included, so I produced a functional equivalent to the noise utils and confirmed I was able to produce the same terrain maps from the new library as the old library. Since this is pure C#, it can be ran on both Mono and .NET without any additional binary dependencies, which wrapping libnoise needed. This will be relevant if I make my terrain generation publically available.

Took some time over the last week to produce Christmas lights in Second Life. I made some last year that worked well, using particle effects, but since many snow, smoke or steam effects use more particles then they should, this only worked in certain conditions. Revamped the system to use some temp-rez objects (which look nicer after some prim tweaking), decreased script load (due to not having a moving object for placing the particles) and made the placement system more sensible. It's now at a point where I feel other people could use it. I also added gravity effects, so the line of lights hangs, but I feel this is going to need more work before I'm happy with it over long spans of lights.

I also continued work on the EvoGrid simulation manager, working on the database design for storing analysis results, while allowing for future expansion and re-evaluation of previously completely evaluated data.