[PATCH] compositor: Disallow negative geometries in backend output configs

[Auke Booij]

On 24 October 2015 at 08:23, Jasper St. Pierre <jstpierre at mecheye.net> wrote:
> I'm struggling to understand the motivation for this patch.
>
> krh has always said that you need to think of uint and int as two
> entirely separate types -- mixing both in math will likely screw up.
> You can see this in other places -- widths are often expressed as
> signed ints in the protocol, not unsigned ints.

Is it maybe time to fix the protocol before it becomes a problem? Not
sure what a sane process towards this would be...