On 08/19/2011 10:58 AM, Dan McCabe wrote:
> On 08/18/2011 01:23 PM, Kenneth Graunke wrote:
>> On 08/18/2011 12:02 PM, Matt Turner wrote:
>>> On Thu, Aug 18, 2011 at 2:38 PM, Eric Anholt<e...@anholt.net> wrote:
>>>> + bool progress = true;
>>>> + while (progress) {
>>>> + progress = false;
>>>> + progress = dead_code_eliminate() || progress;
>>>
>>> || progress is always false, though maybe it's just written like that
>>> to match the style of the code.
>>>
>>> I'd have just written this as
>>>
>>> bool progress;
>>> do {
>>> progress = dead_code_eliminate();
>>> } while (progress)
>>>
>>> Matt
>>
>> The idea is that we'll add more passes later, so eventually, progress
>> won't just be false. Using the do-while loop is probably clearer though.
>
> If that's the case (that we'll add more passes later), this should have
> been noted in the commit msg. It would have prevented people from
> scratching their heads about the logic.
Well...it's basically mirroring the equivalent loops in brw_fs and the
GLSL compiler's do_common_optimization function. So it looked pretty
familiar to me.
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
