<d1b2>
<david.rysk> Could I request a review of 672? @azonenberg
<d1b2>
<johnsel> I'll look at it as well
<d1b2>
<david.rysk> alright. I was told that tests are broken on headless, which this should fix
<d1b2>
<johnsel> well they're broken because it needs Vulkan
<d1b2>
<david.rysk> no, they're broken because the init/teardown is in main() instead of in the callback functions that Catch2 calls at the start and end of the test
<d1b2>
<david.rysk> when that's in the callback functions, they aren't called when Catch2 is merely listing the available tests
<d1b2>
<david.rysk> meaning -- build being broken
<d1b2>
<david.rysk> they won't work without Vulkan, obviously, though they all pass with Lavapipe for me, so that would be one way to run them on CI
<d1b2>
<johnsel> I see well broken in 2 ways then
<d1b2>
<david.rysk> broken both build and run 🙂
<azonenberg>
@johnsel this should speed the build up a bunch and also eliminate those glfw init failed errors we get during catch_discover_tests
<d1b2>
<johnsel> as for the run, we will have Vulkan on the new vm
<azonenberg>
(once it's debugged)
<d1b2>
<david.rysk> yeah, that PR makes it so that catch_discover_tests does not try to init Vulkan anymore
<d1b2>
<david.rysk> anyway, my larger cmake rework is still being worked on, but feel free to test! I think it's reliable on most platforms now.
<d1b2>
<david.rysk> I'm also going to have to modify the documentation
<d1b2>
<david.rysk> also if someone could bump our nativefiledialog-extended to latest upstream that would be helpful
Bird|otherbox has quit [Remote host closed the connection]