On Thu, Sep 8, 2011 at 3:53 PM, Benjamin Franzke <benjaminfran...@googlemail.com> wrote: > 2011/9/8 Chia-I Wu <olva...@gmail.com>: >> On Thu, Sep 8, 2011 at 3:11 PM, Benjamin Franzke >> <benjaminfran...@googlemail.com> wrote: >>> First thanks for taking this on. >>> >>> There are some things I'd like to have addtionally/differently: >>> >>> Supported shm formats are exposed via a "format" event as well >>> (like the supported drm formats), so the config creation logic is the >>> same for drm and shm, and I think it can remain in native_wayland.c >>> >>> We need roundtrips to check that we get at least one supported format. >>> >>> I've attached two patches (heavily based on your last two) that do this, >>> are you ok with using them? >> That is great. Sure. >> >> I've noted two minor issues or typos, fixed by the attached patch. >> One is that param_premultiplied_alpha can be set in the common code >> and only when the display has both HAS_ARGB32 and HAS_PREMUL_ARGB32. >> The other is that we should not claim PIPE_FORMAT_B8G8R8A8_UNORM >> without HAS_ARGB32. If it looks good to you, I will commit an updated >> version of your patches. > > Ok I see the point, though i can imagine compositors may want > to expose only premultilplied argb, which egl users couldnt use then, > but your right, we shouldnt expose alpha at all if one isnt available. > So I'm fine with you commiting that change. Yeah, another restriction coming from EGL_VG_ALPHA_FORMAT_PRE_BIT and how native.h models it. Anyway, we can adapt when there is a need.
-- o...@lunarg.com _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
