On Mon, 04 Sep 2023 at 07:24:55 -0400, Jeremy Bícha wrote: > On Mon, Sep 4, 2023 at 5:00 AM Simon McVittie <[email protected]> wrote: > > I'll apply at least that, but if new architectures can't have llvmpipe > > for a while, then perhaps it would make more sense to apply the mips% > > workaround to every architecture. > > I very recently proposed > https://salsa.debian.org/gnome-team/gtk4/-/merge_requests/17 which > does the same thing as the patch proposed originally in this bug.
I'm testing a build of that, adapted to collect the image diff output for all GSK rendering variants (so that we can debug "gl border-one-rounded flipped") and apply the softpipe workaround unconditionally (because any architecture could in principle be using softpipe). Upstream bug report for the rendering being different with softpipe: https://gitlab.gnome.org/GNOME/gtk/-/issues/6085 smcv
