Sway won't start anymore with sway & wlroots current master branches due to new libseat dependency

[tsdh]

I use sway compiled with a subprojects/wlroots checkout and update sway & wlroots daily (at least on work-days). However, since I did just that today, sway didn't start anymore. The current master commits exhibiting this issue are:

• sway: a558866
• wlroots: b36af22c

Here is the output I get when running sway -dV using the default config.
https://gist.github.com/tsdh/164c2468b27469848c17d3eb4fa407bb

The log indicated that I might want to start using env WLR_LIBINPUT_NO_DEVICES=1 sway -dV instead. So I did and got this output.
https://gist.github.com/tsdh/2af19f08a05e2bb487229372f3e09a3c
With that, sway starts but seems to be dynfunctional, e.g., none of the default key bindings like Mod4+2 to switch to the second workspace do work. Also Ctrl-Alt-Fx to switch back to a console doesn't work. I need to ssh into the computer from another machine and kill sway in order to be able to recover.

Right now, I went back to these two commits in order to use sway again:

• sway: 1a0f86b refactor icon_for_window function
• wlroots: 9f012cac drm: check for PRIME support

Please fill out the following:

• Sway Version: sway version 1.6-1a0f86be (Apr 12 2021, branch 'HEAD')

[emersion]

libseat is now required for logind support.

[tsdh]

Ah, thanks, I've checked out https://github.com/kennylevinsen/seatd into subprojects/seatd as a sibling of wlroots and recompiled. Now it works again.

Feel free to close the issue. Or let it open some days. I guess others will probably get that issue, too.

[markstos]

Before closing, it could be useful to update the title to note "due to new libseat dependency"

[tsdh]

@markstos Makes sense, done!

[emersion]

swaywm/wlroots#2839 should improve error reporting.

[J0nnyMak0]

I just wanted to add that the reason I removed seatd was because it was complaining saying:

00:00:00.003 [wlr] [libseat] [libseat/backend/seatd.c:70] Could not connect to socket /run/seatd.sock: No such file or directory

[emersion]

I think this message should be downgraded to INFO or DEBUG, since it's just probing for seatd availability. cc @kennylevinsen, what do you think?

[markstos]

Just a user here, but I agree. The message looks like it's a problem that /run/seatd.sock is missing, when it's apparently not. It could be clarified " Could not connect to socket /run/seatd.sock: No such file or directory. Optional seatd support disabled` That clarifies that an optional feature has been disabled-- far less worrisome to a user.

…

On Wed, Apr 14, 2021 at 9:42 AM Simon Ser ***@***.***> wrote: I think this message should be downgraded to INFO or DEBUG, since it's just probing for seatd availability. cc @kennylevinsen <https://github.com/kennylevinsen>, what do you think? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#6179 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAGJZKNN33ZZWYVS35JOB3TIWLV5ANCNFSM42ZIQOZQ> .

-- *Mark Stosberg* (he/him) Director of Systems & Security ***@***.*** | 765.277.1916 https://www.rideamigos.com <https://rideamigos.com/> Changing the way the world commutes. <https://www.linkedin.com/company/rideamigos> <https://www.twitter.com/rideamigos> <https://www.facebook.com/rideamigos> <https://www.instagram.com/rideamigos> <https://rideamigos.com/newsletter-sign-up/>

[kennylevinsen]

I think this message should be downgraded to INFO or DEBUG, since it's just probing for seatd availability. cc @kennylevinsen, what do you think?

Hmm, it was upgraded from debug to error due to lacking visibility for users aiming for seatd support. All they saw at the time was the final message that no backend worked. wlroots also used to print errors for each failed backend, so it was in line with that.

Maybe just add an INFO after it failed saying we skip the backend? If feels a little wrong that an error condition would be hidden when the loglevel is set to LIBSEAT_LOG_LEVEL_ERROR, which leads to confusion.

Could not connect to socket /run/seatd.sock: No such file or directory. Optional seatd support disabled

This message is misleading when logind support isn't enabled, which is the case for quite a few distributions and platforms.

[kennylevinsen]

Maybe just add an INFO after it failed saying we skip the backend?

Just realized we already do that.

struct libseat *backend = NULL;
for (const struct named_backend *iter = impls; iter->backend != NULL; iter++) {
	backend = iter->backend->open_seat(listener, data);
	if (backend != NULL) {
		log_infof("Seat opened with backend '%s'", iter->name);
		return backend;
	}
	log_infof("Backend '%s' failed to open seat, skipping", iter->name);
}

[J0nnyMak0]

Just to clarify, I have no issue with the message. I think it is clear. I just meant to say that I removed seatd before it was strictly required. Now that it is required, I think the build should bail if it doesn't find seatd. The build succeeding and then Sway failing to launch is a tad unfriendly IMHO.