> Always transparent is unacceptable. See the performance discussion above.
>
> Also, needs fixing; mir_pixel_format_bgr_888. But ignoring that, I think this
> is overall the wrong approach. We need to at least propose a solution that
> doesn't bind us to being stuck with only one code path which impacts
> performance too much.
>
> If you just need a consistent level of transparency (and later, blur), then
> please reject this proposal and replace it with something that sets the
> surface alpha value instead.
>
> If you need varying transparency then please rewrite this proposal so that it
> does not default to blending always-on.
> Always transparent is unacceptable. See the performance discussion above. format_ bgr_888. But ignoring that, I think this
>
> Also, needs fixing; mir_pixel_
> is overall the wrong approach. We need to at least propose a solution that
> doesn't bind us to being stuck with only one code path which impacts
> performance too much.
>
> If you just need a consistent level of transparency (and later, blur), then
> please reject this proposal and replace it with something that sets the
> surface alpha value instead.
>
> If you need varying transparency then please rewrite this proposal so that it
> does not default to blending always-on.
+1