ZDoom 2.0.98 Bugfixing bugfix release

Post by **Graf Zahl** » Sat Nov 19, 2005 11:02 am

That sounds like it creates an incorrect z-buffer. This sounds suspiciously like a driver bug. I'll add a '-nostencil' option to the next release. If it still screws up with that I'm out of ideas.

Post by **Chris** » Sat Nov 19, 2005 6:53 pm

The Ultimate DooMer wrote:(I get lovely shades of red and green when I try :p)

That seems to happen often when you get Microsoft's software OpenGL driver. Any possibility MS's software driver would get picked over the hardware one when using windowed mode?

Post by **Graf Zahl** » Sun Nov 20, 2005 2:55 am

Sure something like that can happen when the driver is buggy. Windows' GL initialization API is pure shit and crap like this really doesn't surprise me. The driver should at least report that the returned mode isn't accelerated which I suspect is the problem here.

Post by **Chris** » Sun Nov 20, 2005 4:03 am

You can check the GL_VENDOR and/or GL_RENDERER strings (I forget which) for Microsoft's values, and fail if it matches.

Hirogen2 · Post by **Hirogen2** » Sun Nov 20, 2005 4:05 am

I'm happy with the GL libs nvidia provides. Both Windows and Linux.
What's the big deal? After all, the interface is the same, so the user can select at runtime which library to use.

Post by **Graf Zahl** » Sun Nov 20, 2005 4:07 am

Chris wrote:You can check the GL_VENDOR and/or GL_RENDERER strings (I forget which) for Microsoft's values, and fail if it matches.

That library behaves properly. In this case I get a context from the driver but it doesn't behave properly. Normally a driver should indicate if there's some feature requested that isn't supported in hardware.