Patrice Mandin wrote: > Le Thu, 06 Sep 2007 13:22:40 +0100 > Keith Whitwell <[EMAIL PROTECTED]> a écrit: > >> Most likely it is just there for sanity. It's possible that the >> clipper expects this, but some quick testing will show whether this >> is true. Feel free to remove the assert but keep an eye out for >> oddness subsequently. > > It seems you are right, the clipper needs it. We got gears running, but > as soon as a poly is partly out of viewport (clipped), rendering is > completely false. > > Would it be possible to fix the clipper to deal with the POS attribute > not being the first? Or should we have to swap data from vertex buffer > before sending it to the hardware?
The intention is to EOL all the tnl/ code real soon now and move over to the Gallium/softpipe driver model. This will be much better behaved about this type of stuff. For the meantime, I would suggest hacking up whatever is simplest to keep the old code happy for the next few months until Gallium (ie the softpipe branch) is up to speed... Keith ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Mesa3d-dev mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
