Check if gtk is initialised in get_monitors before assuming so in get_monitors

[JAicewizard]

fixes #1941

[jneem]

Do you know why the CI is failing? Should it be gtk::init?

[JAicewizard]

uhh, im not sure how that slipped past me. no its supposed to be GDK, but I needed to scope it first. Maybe a rebase mistake.

[jneem]

Ok, because I think gdk::init exits on failure, whereas gtk::init doesn't. (I'm not sure, because the rust bindings don't seem to include gdk_check_init, and the docs for gdk::init are empty...). I suspect we don't want to exit on failure here, and even if we do, I think we need a better error message before exiting.

[JAicewizard]

Yes gtk::init does not terminate the program on failure, instead it falls back to a textual interface setup presumable without GDK.

If we want to use gdk::DisplayManager we need to init gdk, if thats indirectly via gtk or not we somehow need to init gdk. Using the gtk::init might succeed in that call specifically (with an error), but later it will crash when trying to use gdk::DisplayManager anyways.

[jneem]

I mean if gtk::init gives an error, then we can just print an informative error message and return Vec::new(), right? There's no need to create the gdk::DisplayManager if we know that it will crash...

[jneem]

Oh, I think this fixes #1941, right, not #1800.

[JAicewizard]

I am such a messy person sometimes, yeah you're right.