3 ms·

Tons of X11 infra will have to remain for a long time to support X11-only "legacy" apps that can't easily be updated. There's a long tail of that from games to

by arp242 3y ago

Tons of X11 infra will have to remain for a long time to support X11-only "legacy" apps that can't easily be updated. There's a long tail of that from games to scientific software to just unmaintained software that still works and is used. The major bit will be when and if toolkits like GTK, Qt, SDL, etc. drop X11 support and applications update to X11-less versions. AFAIK there are no plans for that and is still years away (there was some talk about GTK 5, but nothing firm and no one even started work on GTK 5). I also think Wayland will eventually replace all of X11, in the same way as IPv6 will eventually replace all of IPv4.

cies 3y ago
> Tons of X11 infra will have to remain As XWayland. XWayland will be the new X.
- arp242 3y ago
  Well, also libX11 and all of that type of stuff. And as I understand it XWayland shares a lot of code with the regular xserver (it's built from the same repo).
  - cies 3y ago
    Yes, it just sits in a Wayland "container". So we can have security/fractionalScaling/modernMultiMonitor by default and X.oldness when required for that last bit of stubborn software.
    - arp242 3y ago
      What I mostly meant was that for people who prefer to use X11, that will probably remain viable for a long time, without any Wayland. There's this notion that X11 is "deprecated" and that "no one is maintaining it", but the reality of it is far more nuanced. The main "flag day" will probably be when toolkits start shifting away from X11 to Wayland only, and applications start using them (there's usually big lag between release and widespread adoption of new GTK versions and such). Some qualified language in all of the above, because hard to predict the future etc. etc.
      - cies 3y ago
        RedHat pushes hardest, they basically do GNOME, GNOME does GTK. So GTK will be first --my prediction-- to drop X11 support. People that prefer X11 is not what I'd optimize for. To me X11 is deprecated, unmaintain{ed,able}. It became too much of a stumbling block for Linux and other opensource OSes. It's painful, but we have to move on. Wayland is not perfect but it seems to be doing the job lately. In two year --my prediction-- it will be more stable than X11 in all aspects.
        arp242 3y ago
        > People that prefer X11 is not what I'd optimize for. I wouldn't either, but that doesn't mean it won't work. both Wayland and X11 are fairly stable AFAIK. For my part I'd rather not rewrite all my X11-specific stuff. For some I probably need to have direct compositor support as you can do less with scripts, so that's not so easy. I'll probably have to spend the effort at some point, but until I don't have to: why bother?
      - yjftsjthsd-h 3y ago
        > What I mostly meant was that for people who prefer to use X11, that will probably remain viable for a long time, without any Wayland. Or, if necessary, with just a tiny bit of Wayland; XWayland rootful mode lets you run a full X stack with Wayland as little more than a shim to the graphics driver. As a worked example, Puppy Linux implemented this: https://github.com/puppylinux-woof-CE/woof-CE/pull/2265 https://github.com/puppylinux-woof-CE/woof-CE/pull/2265

by arp242 3y ago

cies 3y ago

> Tons of X11 infra will have to remain As XWayland. XWayland will be the new X.

arp242 3y ago
Well, also libX11 and all of that type of stuff. And as I understand it XWayland shares a lot of code with the regular xserver (it's built from the same repo).
- cies 3y ago
  Yes, it just sits in a Wayland "container". So we can have security/fractionalScaling/modernMultiMonitor by default and X.oldness when required for that last bit of stubborn software.
  - arp242 3y ago
    What I mostly meant was that for people who prefer to use X11, that will probably remain viable for a long time, without any Wayland. There's this notion that X11 is "deprecated" and that "no one is maintaining it", but the reality of it is far more nuanced. The main "flag day" will probably be when toolkits start shifting away from X11 to Wayland only, and applications start using them (there's usually big lag between release and widespread adoption of new GTK versions and such). Some qualified language in all of the above, because hard to predict the future etc. etc.
    - cies 3y ago
      RedHat pushes hardest, they basically do GNOME, GNOME does GTK. So GTK will be first --my prediction-- to drop X11 support. People that prefer X11 is not what I'd optimize for. To me X11 is deprecated, unmaintain{ed,able}. It became too much of a stumbling block for Linux and other opensource OSes. It's painful, but we have to move on. Wayland is not perfect but it seems to be doing the job lately. In two year --my prediction-- it will be more stable than X11 in all aspects.
      - arp242 3y ago
        > People that prefer X11 is not what I'd optimize for. I wouldn't either, but that doesn't mean it won't work. both Wayland and X11 are fairly stable AFAIK. For my part I'd rather not rewrite all my X11-specific stuff. For some I probably need to have direct compositor support as you can do less with scripts, so that's not so easy. I'll probably have to spend the effort at some point, but until I don't have to: why bother?
    - yjftsjthsd-h 3y ago
      > What I mostly meant was that for people who prefer to use X11, that will probably remain viable for a long time, without any Wayland. Or, if necessary, with just a tiny bit of Wayland; XWayland rootful mode lets you run a full X stack with Wayland as little more than a shim to the graphics driver. As a worked example, Puppy Linux implemented this: https://github.com/puppylinux-woof-CE/woof-CE/pull/2265 https://github.com/puppylinux-woof-CE/woof-CE/pull/2265