chjj
Tabbed windows become completely invisible after switching the monitor outputs
Platform: Arch Linux, WM: i3 version 4.14.1 GPU, drivers, and screen setup: Intel HD3000, xf86-video-intel - 2.99.917, two monitors configured with xrandr Compton version: v0.1_beta2-87-g316eac0 installed from AUR, package: compton-git
Compton configuration: default Airblader's configuration without any changes https://github.com/Airblader/dotfiles-manjaro/blob/master/.compton.conf
Steps of reproduction
Usually when switching from dual monitor setup to internal screen i used to completely reload my i3. As part of reloading procedure compton is also reloading every time like this:
killall -q compton
while pgrep -x compton >/dev/null; do sleep 1; done
compton -b
If there were any tabbed windows on any of my workspaces, after monitor configuration has changed all of them look like this(completely invisible):
Reloading compton(as described above) doesn't change anything, windows are still invisible.
I've tested this with i3-gaps and previous version of i3 with the same results.
Expected behavior
All windows are visible.
Current Behavior & Other details
I've figured out how to fix this, but this seems overkill to me for just switching the monitor outputs.
Using xprop one should remove the window property:
xprop -id <window id> -remove _NET_WM_WINDOW_OPACITY
After this window goes back to normal again.
Is there a better way to do this?
Those lines setting opacity rule for windows of state _NET_WM_STATE_HIDDEN might be a cause of the problem. But if you remove those lines, you will see tabbed windows overlapping on top of each others..
@patzm Might be worth trying this with the newer fork of compton (picom) & if you still have the problem pop an issue in over there
Oh, I actually was already using picom for a while now. I see that compton links to picom. I installed the arch community version of the package.
I will now try with the experimental backends, which I didn't do before
compton -b -d :0 --experimental-backends
With that, I guess my contribution to this thread was a bit off, as it is (at least to some degree) a different library. Let's continue discussing it over there. Thanks @06kellyjac for the pointer.