Drop our logind code, rely on seatd instead.
cc @kennylevinsen
I'm not sure this is a good idea, given that seatd calls itself "alpha" right now.
Can we hold off on this until it matures more?
seatd calls itself "alpha" right now.
https://git.sr.ht/~kennylevinsen/seatd/commit/0d855a28f20ab3d7b2058fbe58a172a97affbde5
@kennylevinsen Do you consider it stable to rely on?
I'm not too worried about the libseat code.
Why not?
It's pretty similar to wlroots' code, and has been around for a while.
I guess the next step would be to make libseat a required dependency for wlroots, and nuke our direct session (and perhaps our session interface).
and nuke our direct session
why? I think it is useful, for simplicity's sake.
Because it's code which I'd rather share with other compositors than maintain, just like e.g. libinput. It's pretty trivial to drop seatd into subprojects/ and make Meson build it as a subproject anyways.
libseat is now required and replaces all of our session code.
If you were using the direct backend and libseat is not yet packaged for your distribution, the replacement is now:
subprojects/seatd-Dseatd:builtin=enabledAnd things should work just like before.
Most helpful comment
https://git.sr.ht/~kennylevinsen/seatd/commit/0d855a28f20ab3d7b2058fbe58a172a97affbde5