wayland: Update to 1.22.0

[Chubercik]

https://gitlab.freedesktop.org/wayland/wayland/-/releases/1.22.0

[akien-mga]

Thanks for looking into this.

Some API changes require downstream changes too: https://github.com/godotengine/godot/pull/87977/files#diff-c86c10b4f67151b187e2f364912792198ba5a3ef23346cfe43cb054b466b4c59

[Riteo]

Yup, we need to implement even just stub event handlers. Note that this isn't terribly important yet as we don't use all the new features, which is on my TODO list.

Edit: to be clear, for now the new handlers must be stubs, as we're purposely asking for older versions. It's not needed to implement the various logic yet. If that's really needed though, please let me know.

[Chubercik]

At first I added short implementations that looked like this:

void WaylandThread::_wl_surface_on_preferred_buffer_scale(void *data, struct wl_surface *wl_surface, int32_t factor) {
	WindowState *ws = (WindowState *)data;
	ERR_FAIL_NULL(ws);

	ws->preferred_fractional_scale = factor;

	DEBUG_LOG_WAYLAND_THREAD(vformat("Window preferred buffer scale %d.\n", factor));
}

But given that this is way out of the scope I was prepared to tackle in this PR, I'm leaving them empty for the time being.

[Chubercik]

Does this:

void WaylandThread::_wl_surface_on_preferred_buffer_scale(void *data, struct wl_surface *wl_surface, int32_t factor) {
}

fall under the definition of "stub" @Riteo? If so, I can rebase the PR and it'll be ready for review :)

[Riteo]

@Chubercik

But given that this is way out of the scope I was prepared to tackle in this PR, I'm leaving them empty for the time being.

Yeah, it's best to keep them empty, as this code won't be run and thus can't even be tested yet. Also, we can't just randomly bump the version as there's a lot more event handlers or changes between versions that we have to keep in mind. This is all planned but for now I'm concentrating on other important foundational stuff now that most of the work is merged.

fall under the definition of "stub" @Riteo? If so, I can rebase the PR and it'll be ready for review :)

Yup!

FTR, there's also another approach, which is to assign a simple void "noop" method, but I think that C++ might annoy us (it's way less lenient with pointer assignment) and we'd still have to implement those anyways later so IMO it's more worth it to make empty stubs as you've done, which is what I'd recommend :D

[Riteo]

Looks fine, thanks!

[akien-mga]

Thanks!