On Sat, 13 Apr 2002, José Fonseca wrote: [snip] > BTW Leif, how are going the glean tests of mach64? I noticed you commited > a patch, but I ran glean on mach64 and there still are a bunch of errors > in blendFunc and logicOp. Since you're more in the subject could you give > me a brief update of the remaining problems and what may be causing them > so I can help?
I haven't made much progress yet. I got sidetracked with 24-bit depth and the depth buffer. I was able to allocate only enough offsreen mem for a 16-bit depth buffer by passing a smaller "height" to the offscreen allocation function, but I haven't commited anything yet. I also noticed that the draw engine doesn't seem to write to the alpha channel at 24-bit depth, although WritePixels and ReadPixels work. I have the necessary changes to "disable" the alpha channel in the glx visual and ReadPixels in my tree, but I'd like to resolve the ReadPixels problems before committing it. So the problem is still that ReadPixels returns zeros for the color channels on many of the tests, even though I can see drawing happening. Most of my checkins have been stuff that was in my tree for a while and I wanted to sync up the branch. I've been looking into timing of the read and state/context changes that could affect the read buffer offset as potential problems, but I haven't found anything yet. One reason I suspect read buffer offsets is that the texenv Mesa demo, when run without -db (double-buffering), has old textures or framebuffer images showing through in the textures using software fallback. I was thinking that perhaps Mesa was reading from the wrong place in the framebuffer to do alpha blending for the fallbacks. -- Leif Delgass http://www.retinalburn.net _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel