On 08/28/2011 09:45 PM, Eric Anholt wrote: > Ken didn't like how I was handling new instruction generation for the > array stuff, and it convinced me to try a cleanup I had been thinking > about. I think this ended up nicer than before. > > I don't like the extra set of parens per emit() line (back to how the > FS used to be, where I mis-balanced parens pretty regularly), but that > doesn't seem really avoidable in this change while not always emitting > to the end. The reordering didn't really bother me, though. > > Not all instructions are converted. For example, ENDIF, ELSE, DO, > WHILE, each of which are used once and take no operands, so they don't > seem worth it. Also, SEL can be either predicated or have embedded > comparison, and I couldn't decide what I wanted to do with it, after > CMP worked out so well.
For the series: Reviewed-by: Kenneth Graunke <kenn...@whitecape.org> This is more than I was thinking, but does seem nice. CMP did work out really nicely. _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
