GTK: Consider dropping the X11 back end | Hacker News Reader

GTK: Consider dropping the X11 back end

gitlab.gnome.org

42 pointsCloudef45 comments

Well I still use GNOME X11 on my desktop; every time I try to switch to GNOME Wayland I eventually encounter some issue which forces me to switch back. I usually encounter some game which works under X11 but refuses to launch under Wayland with Steam Proton, and suspend is still basically completely broken with an nvidia card (and my GPU is too old to ever be able to use the new open-source kernel driver). I've also experienced much worse performance and more input lag in certain games under Wayland than under X11.

Maybe all these issues will eventually be fixed and I can comfortably switch to Wayland. But if GTK drops X11 support before these showstopping bugs are fixed, I will have to stop upgrading GTK.

EDIT: To be clear, this isn't meant to be a dig at Wayland; I've used it on laptops with Intel and AMD GPUs and it works great, much better than X11 even. I bet most of my issues are related to nvidia. I wouldn't be surprised if it's impossible to fix these problems without fixing nvidia's proprietary driver. But as a user, I don't care. My 5 year old 1080Ti still plays all the games I need it to (under X11) and I'm not looking to replace it any time soon.

I wouldn't be surprised if this is 6 years away. Nvidia will work by then.

Maybe, maybe not. Ideally, all bugs everyone has with wayland in all configurations will be universally fixed in a few years. But I really wouldn't be surprised if nvidia-specific issues will only be fixed when using the new open-source kernel module they released some time ago, which means all 10-series cards and earlier won't see any improvements. (I also wouldn't be surprised if these issues don't get fixed at all, since wayland has been here for 13 years and still has these major issues.)

Wayland has been used by actual people for less than 6 years. Fedora 25 released with it by default in late 2016. Before that it was heavily in development.

This is why software originally started having telemetry. This is a decision that is quite easily answered by telemetry: You bet that a good, say, at least 30% of users are definitely sticking with X11. In any server setting, where the maximum youll ever do is VNC into it or hook up a display via VGA, or any setting where actual real work is being done (custom software, etc.) its likely that X11 is even a requirement (correct me if im wrong).

Its like deprecating support for windows 7; sure nobody should be using it, but you bet there is a huge (absolutely speaking) number of people using it still, for one reason or another.

To clarify:

- VGA has nothing to do with it, but when I do sysadmin stuff and just walk in, turn on the monitor, i dont want to first troubleshoot wayland, or even install it in the first place

- "real work" is the wrong way to put it: i mean when a company has written an application specifically for their own usecase, and (if theyre unlucky) targeting X11

By the time that the vast majority of users move to Wayland, it will be quite obvious that any X11 usage has declined, purely due to the discussions on the community grapevine attesting to that fact.

To decide whether a platform needs to be dropped, you don't need anything as fine-grained as telemetry. If you want to be conservative in your estimates (which, if you care about backwards compatibility, you should be), then it will become quite obvious that no-one is using a particular platform any more.

Telemetry won't help, the issue is that there are few developers contributing to X11 anymore. The project is on the way out. Ninety-nine percent of users could be using it and it wouldn't change that. It's not like Windows 7 at all, Microsoft can re-allocate engineers to work on that if need be. In open source there is no manager who can re-allocate anybody.

I am not able to share screen while on wayland session. Sadly i really need that functionality for work so i use x, that is the only issue i have with wayland. Good news is that they only plan to do this in gtk5 hopefully things will get better until then.

From what I see, most Wayland aware applications have converged on using PipeWire (through XDG Portal) for screen sharing.

Personally I use PipeWire anyway but I'll admit telling people to replace more of their setup as a solution isn't very nice.

Also there might be oddballs like how Zoom used to use the GNOME screenshot API for a while.

Latest Zoom release supports Wayland screen sharing.

The only oddball these days is Discord, because they refuse to ship an updated version of Electron. Works fine in-browser.

You can absolutely run pipewire for screen sharing while running something else for an audio server. If your distribution doesn't provide this as a supported setup, I suggest you check with them.

On Fedora 36 with Gnome/Wayland screen sharing works (using RDP protocol).

Works well, or just barely? Because the last time I checked it still had highly annoying 500ms lag and would sometimes miss screen updates.

They are delusional. I can see this being a legitimate move after X11 support dwindles to low double-digits, but it's far from being the case.

There are far more pressing issues, like the thumbnail file picker (or having a decent file picker at all, actually). They should focus on those issues.

Well GTK sucks almost as much as Wayland, so they're welcome to each other. The GTK file picker has to be most annoying thing. It has numerous horrible bugs that are if ignored for 5+ years. You cant choose any view except filename list, so if you work with media and are used to big thumbnails, forget Gnome.

And X11 now fixed the couple of issues that Wayland people complained about most.

The funny thing about picker is that GTK supports alternative filepickers on flatpak offically, but non-flatpak is considered an "debug feature" [1]

1: https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/4829

Try to understand what you comment on.

Users blindly used this env var and break their system. Inside flatpak it's required to work but outside it could be running in a process that shouldn't be using the portal for various features.

There are also various xdg-desktop-portal bugs that need to be fixed there first. Like some portals fail on the host.

Because it is a debug feature, that environment variable affects far more than the file chooser.

The actual problem is that it is not part of the FileChooserNative API to use the XDG Portal API outside of Flatpak. I just opened an issue for this specific usecase: https://gitlab.gnome.org/GNOME/gtk/-/issues/5026

>The way to do that is to install your applications as flatpaks.

Yes absolutely helpful as always

> And X11 now fixed the couple of issues that Wayland people complained about most.

No, they did not. The tearing is still there.

Haven't seen tearing in X11 for years

Well, I always had extremely bad tearing with X11 and Intel, AMD GPUs, and no issues with Wayland.

Try a monitor in rotated (vertical) mode.

https://linuxreviews.org/HOWTO_fix_screen_tearing

>Since this issue found its way on Phoronix and reddit I'm going to temporarily lock it, to ensure a modicum of decency and avoid lowering the bar of the comments.

Classic gnome dev's ;)

Classic serious open-source project. Nothing worse than having a serious internal discussion overtaken by a flood of hundreds of meme comments or angry posts from people without the necessary knowledge to participate in the conversation.

Professional like that?

https://news.ycombinator.com/item?id=27075980

Gnome is more like woke, anti other DE's. Professional is probably the least attribute i would attest towards gnome.

The comments are kind of hilarious: You can't speak of Wayland's issues as reasons to not do this and then the discussion becomes "well we are just a client anyway". If you just want to do something, why pretend to have a discussion? It's just frustrating for the reader - just drop it so everyone can either show you that they agree or move on from what you are maintaining.

There are just too many things I use that are very much dependent on X.

Mostly what comes to mind is Synergy and ThinLinc.

GNOME devs seem dead set on killing everything that works, but they don't approve of. [1]

[1]: https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/4829

It's in everyone's interests for maintenance of legacy stuff to be as easy as possible.

That means building stuff with clear boundaries and simple API's and the ability to split parts off to run in sandboxes or containers.

Build it like that in the first place, and you'll be able to maintain compatibility for decades, even with few users and minimal development effort.

I suspect none of the above is true in this case.

Funny you mention that. X11 has been maintained for decades.

Tell me, does this Wayland improve the User experience? Or merely the developers?

It improves user experience. Screen tearing and otherwise displaying partially updated frames is a bad user experience.

This will cause another distro-split and distro hopping migration for many Linux Desktop users who see problems in apps that are running in Wayland. The whole problem of defining Desktop Linux support is just about to get even worse.

As if there's no myriad of distros these days? As someone that wants a modern desktop with unified foundations, good riddance. I'll gladly use any GNOME+systemd distributions.

The reason there's half a billion distributions is not because what's mainstream is bad, but because it's so easy to create a new distribution any bikeshedding argument results in a split. And people just _love_ to bikeshed.

They broke gnome extensions in release 42. Not very happy about it!

Would be nice to know what is the issue with supporting x11 backend.

No GTK developer uses X11. Wayland has been the default for many years. So what happens is just bugs pile up, no contributors step up to help, and here we are. These are just future plans for an api version that doesn't exist yet.

Thanks!

It's a lot of work that they don't want to do.

Does this mean that applications using GTK would no longer work on all the desktop environments that don’t support Wayland?

They want force people to adopt Wayland no matter what.

They have zero care about user's needs.

Everybody knows Wayland is far from being a full replacement for X but they obviously don't care.

It's really funny to see how hypocritical GTK/Gnome devs are when they talk about inclusion and diversity because they are the first to exclude people.

They simple refuse to talk about Wayland's problems and try to shut down every discussion.

Indeed, they are the most toxic, and abrasive OSS community so far. GNOME ecosystem went from 3000+ developers to low hundreds, and around ~30 active core maintainers. The most spectacular OSS failure story.

X11 should always be supported

I hope they do, so they can become a minority, i'm sick of these programs with CSD

Reply on news.ycombinator.com