 "Today:..." <- I think `SurfaceInputDispatcher` might be a wrong abstraction entirely? And something more directly `wl_seat`-like might be a better match.
[01:35] <RAOF> Hm. We could switch over to using libepoxy. Sometime….
 "I think `SurfaceInputDispatcher`..." <- That's really interesting. I've never thought about the complexity of this. I've a walcom and both pencil and mouse can interact together. The mouse tries to ghost itself when the pencil is near the toouchpad surface. What a challenge for  developers and testing. Thanks
[07:17] <RAOF> Why? Why? Why do you work on nouveau but not amdgpu?
[08:04] <RAOF> Today:
[08:04] <RAOF> * Engaged in the [dmabuf import bug](https://github.com/MirServer/mir/issues/2958)* 
[08:04] <RAOF>  * Today:
[08:04] <RAOF> - Engaged in the [dmabuf import bug](https://github.com/MirServer/mir/issues/2958)*
[08:04] <RAOF>  * Today:
[08:04] <RAOF> - Engaged in the [dmabuf import bug](https://github.com/MirServer/mir/issues/2958)
[08:08] <RAOF>  * Today:... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/47bfecdb878308813f7571d2d9c2cc8c7660f0c0>)
[15:40] <Saviq> RFC: Migrating our performance tests to Checkbox, I'm thinking to get rid of the explicit `--gtest_filter`s we have (`-Hosted*:*Xwayland.windowed`) that work around failures on `eglstream-kms`.
[15:40] <Saviq> I could ask udev for the driver in use and quirk the failing tests away in the test harness… any other ideas?
[15:40] <Saviq> This will, of course, be impacted by the hybrid, where the decision will be trickier - and we won't yet have a way to even ask Mir to render on one, but output on another, platform?
[15:43] <Saviq> And with that…... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/f0ecc78b37d20d7c07ca2b01919811fddf660925>)
[15:43] <alan_g[m]> What is the advantage of changing a gtest_filter to a nastly hack in the test?
[15:45] <alan_g[m]> Tests should not second-guess whether they really test what they say
[15:48] <Saviq> Didn't mean the test itself, but the test runner (i.e. checkbox-mir). Rather than hardcoding the filter for a number of hardwares in Jenkins, it could skip them based on the environment it's running in
[15:49] <Saviq> Wouldn't be much different to an explicit filter baked into the runner, like it is today
[15:49] <alan_g[m]> That makes more sense
[17:01] <alan_g[m]> Today:... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/939718aa2119f2f4291e8cdbb2803ca710afd309>)