The Tyan K8WE motherboard in hinge has always seemed to have flaky DIMM slots. I seemed to have to wiggle the DIMMs just right in order for the machine to recognize all of the memory. A recent graphics card upgrade nudged things and I was unable to nudge them back into an agreeable position. The board had about a month left on its three year warranty; so, back to Tyan it went.
So I am without my primary development machine for at least a week. It’s hard to believe I’ve had this thing for nearly three years. Last year I upgraded the CPUs to Opteron 285s from the 242s I originally installed; but still it doesn’t seem that old. Its predecessor was a dual Pentium 3 machine that lasted me five years
When hinge is put back together, I think I’ll go ahead and release OpenVRML 0.17.6. I still haven’t fixed that damn JPEG bug; but I think there have been enough less consequential changes to warrant a release.
Another change I’d hoped to work into a point release in the 0.17 series is the use of D-Bus to replace the existing IPC mechanism used in openvrml-xembed. Aside from being generally cleaner and nicer than the existing hacky IPC solution, I’m thinking that expressing the IPC interface in D-Bus should lead to an improved code organization that will in turn make it easier to convert the existing multiprocess approach into a multithread approach for Mac OS X and Windows. This is becoming more pertinent as it seems there may be folks interested in contributing code to a Windows plug-in.