@pixelate @tyrylu @pvagner @elementary ehh that honestly depends on the tool, in general maintainers aren't really in favor I don't think

Well, depends on how big a fix that is - a few missing labels are definitely okay, a huge refactoring because at the end you can fix some small nidpick, I'd doubt it. But you'd at least test it. And of course, cross environment issues (so things that Kde, Gnome and others have to agree upon) need a much more complex discussions.

yeah, in the end it depends on how big the change you're submitting is, because if it's gonna be a few hundred lines of code split across a bunch of files, sometimes entirely unrelated files, then maintainers are right to scrutinize that deeply, because that path leads to unmaintainable code down the line. But yeah, I feel like nowadays, the most important things for accessibility to not only be good, but actually awesome, on linux to be done are in the wayland protocols and AT integration department, rather than anything specific in one desktop environment or another. You know, for example, stuff that would be needed for eye tracking, a proper way of querying the mouse pointer, getting touch gestures and so on, all with a properly sandboxed API. However, that's the hardest part, not because it'd take a lot of code, but because so many people would have to be involved in that discussion, the bikeshedding would probably be monstrous, but I suppose that'll have to be done eventually even so

@esoteric_programmer @tyrylu @pvagner @elementary I shoot that reminds me, I have a laptop with a touch screen so I can like start working on that. It'd all be AI generated code though.

well nope, first we have to look at what exists and try to make this into an actually usable interface which can be used for more things, for example portals are a good abstraction layer. Right now, the fixes which fixed most of wayland accessibility for us are two dbus interfaces the gnome, and kde now I suppose, shell implements, but that's a hack more than an actual solution. That's why I said that at this point, we can't fix that with code because it's not really code that's the problem, it's trying to make something work so that we don't slide into more technical debt than we already are. First things first, I think someone, if not us, should revive the newton branch and make that workable again, then put effort into upstreaming that, because that already has some nice wayland compositor primitives which would help us later. Then, we need an accessibility portal. As a quick hack, sure, you can probably create patches to orca, at-spi2-core, libei, mutter and gtk4 probably, with or without AI, but those would be hacks. But here's a thing, we can brainstorm ideas and see what can be done, do you have a matrix or something? I think discussing the whole thing in this thread would make it too long for most clients to load it :p. After we got something, we can probably talk with more wayland and portal devs, because right now, I was trying but from a position of seeing what sticks on the wall and nothing concrete was done because of that. I can probably try to pull up some of my earlier messages in various rooms about this problem and what I came up with, so I don't reiterate from scratch.