mstoeckl

When I ran this branch of swaybg on an output with 2x scaling, `swaybg -c '#ff00ff'` failed to display a background; see following partial log with WAYLAND_DEBUG=1. ``` [ 76026.226]...

Also, a minor issue, not necessarily worth fixing: every time the output is resized, swaybg creates a new pair of `wl_buffer` and `wl_viewport` objects, and then promptly destroys them after...

What command, exactly, are you using? Also, what cursor theme, and which Wayland compositor? I can't reproduce this myself; however, if I make a screenshot with `grim -g "$(slurp -d)"`...

Another issue with the seat's selection serial value is that the serial used to perform the validation (now `seat->selection_serial`) is shared between all programs using the seat, so if one...

Right, I think I understand it now. The "correct" validation for primary/gtk selection serials is to verify that 1. The client provided a serial value that the compositor previously sent...

I have a rough implementation of the ring-buffer approach; see https://github.com/mstoeckl/wlroots/commits/serial-validation for details. It currently stores the validation ring buffer with the seat, and only validates selections, so there's quite...

> [main.c:560] Missing a required Wayland interface How are you running swaybg? From a terminal, or is it automatically launched by `sway` because of an `output ... bg ....` command,...

What error message is provided when swaybg (that is, the build with more detailed error messages) is run from the config file, in the same way the produced the `[main.c:560]...

> `[main.c:521] Missing a required Wayland interface` Based on the line number in the error message, it looks like you built from the mstoeckl/master branch, not mstoeckl/better-startup-errors . Try downloading...

> `2022-01-30 15:34:00 - [main.c:577] Failed to roundtrip with the compositor` This implies that swaybg is connecting to something, but the connection is broken before it can do anything. Unfortunately,...