Correctly activate windows even if they're not in the iconic state

[Stebalien]

Handle the case where a window is hidden but not iconic. This code would previously error in some cases because a buffer wouldn't have an active window even when it was not in the iconic state, likely due to a focus race somewhere.

fixes #28

[Stebalien]

Unfortunately this patch is more of a fix for the symptom rather than the underlying problem (I think), but it's still more robust than the previous code so I'm reasonably happy with it.

I'm also not able to reliably reproduce this problem, so I'll have to rely on the community to test it.

[minad]

Thanks! I was about to suggest a similar patch. The problem is that the iconic state gets out of sync with the buffer visibility. I think it is better to avoid maintaining "duplicate" state which needs to stay in sync. Maybe it is possible to eliminate some state?

[Stebalien]

Unfortunately, we actually have three variables here:

1. Buffer visibility.
2. Buffer-local exwm-state.
3. The WM_STATE window property.

The exwm-state local variable is just a mirror of the WM_STATE window property (likely for performance?).

I think we need to continue using exwm-state, but only in exwm-layout--auto-iconify? Otherwise, yeah, we should just be checking if the buffer is actually visible or not.

[Stebalien]

E.g., I think the following need to change:

exwm/exwm-layout.el

Line 157 in bf92108

(unless (or (exwm-layout--iconic-state-p)

exwm/exwm-manage.el

Line 747 in bf92108

(if (exwm-layout--iconic-state-p)

exwm/exwm-input.el

Line 360 in bf92108

(when (exwm-layout--iconic-state-p)

[minad]

Okay, let's keep this state duplication problem in mind. We don't have to address this now. From my side this PR looks good conceptually, but I haven't tested it yet.

[Stebalien]

This patch doesn't cause any problems for me, but I'd still like some feedback from the community to see if it actually fixes any problems. I only ran into it once or twice and can't reliably reproduce it.

[rayslava]

It seems that the change helps with focus issue, when some buffers did switch randomly, didn't help with exwm-xim though

Upd: no, the focus issue is not fixed as well :(

[Stebalien]

Ok, I think your bug is https://bbs.archlinux.org/viewtopic.php?id=294301.

I'm going to merge this PR as a fix for my issue (that appears to be unrelated).