Re: [AD] CPU capabilities patch

[ Thread Index | Date Index | More lists.liballeg.org/allegro-developers Archives ]


> And cpu_sse.

No, cpu_sse isn't part of the 3.12 API.

> Yes, we've discussed the issue before I wrote the patch, and
> concluded that the number of affected programs is minimal. The
> changes will also be very small.

Yes, but given that we struggled many times in order not to break the API,
including using convoluted macro trickery, I think if would be stupid to
break it because of the cpu_* variables.

> We could perhaps document this - have a list of API
> changes in betweem 3.x and 4.0, (which will be referenced by just about
> every doc that comes with Allegro ;) so the user can change his program
> accordingly.

Ideally it would contain no API breakage.

> Well some things, such as writting to those variables, will be a bit
> tricky.

Were I a Jesuit (French expression), I would say that these cpu_* variables
were semantically read-only anyway, so we had a right to throw them away
and replace them by cpu_capabilities and macros. Nevertheless, you're
obviously right...

> #define cpu_mmx (cpu_capabilities & CPU_MMX)

Yes. I think this is both necessary and sufficient.

--
Eric Botcazou
ebotcazou@xxxxxxxxxx



Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/