Bartosz Taudul
Bartosz Taudul
Fixed by #1078
This may be relevant: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30014
The slow behavior I described in my application happened when I had no render loop and was doing `xdg_surface_listener.configure` (resize event) -> create/destroy swapchain -> render. I have now switched...
VS Code has the functionality to format only the selection.
What's the failure mode that these changes are protecting against?
CI setup won't be changed.
Thread naming is something you want to have universally, and otherwise you would need to have two systems doing the same thing.
> it is happening consistently; [@wolfpld](https://github.com/wolfpld) ideas? Queue block allocation cost is amortized.
Tracy uses OpenGL to render the window contents. If you want software rendering you need an OpenGL driver that will be able to do it.
It would be helpful if you'd describe changes made in `TracyWorker` and `TracyEvent` (i.e., how the data storage is changed, why, how it's better; step by step).