FWIW, thinking about it a bit more, my opinion is that it might be best to add both a flag to respect scissors and also explicitly pass the MRT buffer/component writemask as well to clear().
Drivers can probably disable/enable scissors and change writemasks, if necessary, in a faster way than the state tracker could (especially given the writemask is currently hidden in the blend CSO, ouch). Another option is to just make it unconditionally respect the current scissor and writemask. That's for Keith/Brian to decide though, of course. BTW, thanks for tackling this, we have one less problem now :) _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
