GNOME doesn't care if wayland is at feature parity. They can require it for GNOME by fiat. They also don't need to care about supporting multiple compositors; they can require all GNOME users to use theirs.

It's not really a "fiat". There just aren't any GNOME developers left that are interested in maintaining the X11 parts of the stack. The Mutter developers are actively working on making the remaining X11 parts optional. Xorg itself is not getting any new extensions, all feature work on display protocols relevant to Mutter is currently happening in Wayland extensions and in Mesa. The major problems left there that involve X are related to XWayland. The GTK developers already redesigned the whole client back end to match Wayland. The other pieces of GNOME have long since moved into protocol-agnostic dbus interfaces. What incentive do they have left to maintain X11 support? There just isn't much there besides legacy compatibility.

Perhaps "fiat" was too loaded of a term. What I mean to say is that, since GNOME is an integrated system, there is no particular need for it to depend on Wayland implementing some random feature that X11 has. On top of that, since so much of Wayland is implemented in the compositor, it's likely that Mutter can make up for any missing features in Wayland.

Therefore, as you say, there's no incentive for the GNOME developers to support X11, and the devs are free to say "Use GNOME on wayland, implement an X11 backend for GNOME/GTK yourself, or don't use GNOME" which, in practice, boils down to "Use GNOME on Wayland, or don't use GNOME" and is, effectively, a fiat to use Wayland if you want to continue using newer versions of GNOME.

I also didn't intend this to be judgemental; GNOME is almost certainly the most complete desktop environment for linux. I personally use KDE as my desktop, but most applications I use are not part of KDE (or even built using Qt). Conversely, a typical desktop user could probably get by using just GNOME applications (plus maybe libre-office for opening MS documents that can't be handled within GNOME). IMO a large part of what makes this possible is by being judicious in what is supported.

>and is, effectively, a fiat to use Wayland if you want to continue using newer versions of GNOME.

Ok, I get what you're saying, but this is not different from looking at GNOME 2 (which only ran on X11) and saying they declared a fiat that you have to use X11. Any project gets to pick its dependencies.

Both of you are concentrating on GNOME DE. GNOME DE doesn't matter. Gtk3/4/5 matters and it, and Gtk using devs writting applications, don't have all the vertical integration and control that the GNOME desktop environment has.

Maybe I'm wrong, but I get the sense that GNOME is the tail that wags the Gtk dog at this point.

You're not wrong about control of Gtk. But my point was that if GNOME does drop such X11 functionality while the GNOME DE may not suffer, the far more important and larger set of applications that use Gtk will. GNOME DE won't have problems but the Linux ecosystem will.

>You're not wrong about control of Gtk

No, the parent comment is wrong. There is no "vertical integration" or control. It's just an open source project. If anyone was interested in maintaining the X11 support, it would happen, or at minimum it would happen in a fork, but there is currently nobody.

If someone is worrying about other parts suffering, they should start contributing now. I doubt this will happen though, nobody has turned up over the past several years, which is exactly why they consider dropping it. Honestly I think most developers who even bother with GTK4 are probably fully on board with wayland, the ones who aren't are still trying to use GTK2 and GTK3.

After looking at gtk4 recently, I also learned it's pretty crippled in some areas as compared to gtk3. I understand their rationale for this (removing things like GtkPlug/GtkSocket, and making it impossible to implement yourself), but I can't say I agree with it. I've started a couple new gtk3-based projects recently, and I don't think I'd ever migrate them to gtk4.

GtkPlug was a pretty busted API that only ever worked on X11, and even then it didn't totally work right. To make it work on other platforms (or allow users to implement it themselves) would require exposing so many hooks in the toolkit that it's not worth it.

Gnome will continue supporting X apps via xWayland. GTK3 and GTK4 apps will always work on both. Anyone building a new app should not be writing for X at this point (its deprecated) so there is no need for X support in a future toolkit.

> Anyone building a new app should not be writing for X at this point (its deprecated).

Did I miss an announcement or something? Last time I checked X was not deprecated.

How would you even deprecate a protocol? Is HTTP/1.1 also deprecated?

I mean I guess you deprecate a protocol by having the implementations tell people to stop using it, then they stop supporting it.

SSLv2 and SSLv3 are surely deprecated protocols, no?

>> Did I miss an announcement or something?

Apparently.

>> Last time I checked X was not deprecated.

Well now you know. Feel free to check again.

If X11 is well and truly on the way out by then, I'll be all for GTK5 dropping X11 support. It doesn't seem like much of the world beyond Gnome is on Wayland but GTK is everywhere. If that remains the case, a Wayland only GTK (that didn't accept X11 patches) would probably result in a massive fork.

I generally use Gnome on my Linux laptop. I often find myself switching it to be Gnome on X rather than Gnome on Wayland. This isn't based on any explicit preference (I genuinely don't care), but I usually encounter some glitchiness that goes away when I take Wayland out of the picture.