bindsym --release behavior different from sway's

[dawsers]

Since 8907f49, scroll's bindsym --release behavior is different than sway's.

In sway, a release key binding triggers when you release the main key associated to the binding (non-modifier key). In scroll, now you need to release every key of the binding, in any order. This lets you do things like the following temporary overview mode:

     mode "overview" {
         bindsym --release $mod scale_workspace overview; mode "default"
    
         bindsym --no-repeat $mod+$left focus left
         bindsym --no-repeat $mod+$down focus down
         bindsym --no-repeat $mod+$up focus up
         bindsym --no-repeat $mod+$right focus right
         bindsym --no-repeat $mod+home focus beginning
         bindsym --no-repeat $mod+end focus end
     }
     bindsym --no-repeat $mod+tab scale_workspace overview; mode "overview"

You press $mode+tab to enter overview mode. While keeping $mod pressed, you use the arrow keys to change focus, and as soon as you release $mod, you will exit overview mode. With this feature you can create workflows similar to other desktop environments, Atl+Tab, etc.

[rempas]

This is awesome! Thank you!

[larskotthoff]

This doesn't always fire. I've tried to reproduce it, but there doesn't seem to be any consistency to it. Nothing in the logs.

[dawsers]

This doesn't always fire. I've tried to reproduce it, but there doesn't seem to be any consistency to it. Nothing in the logs.

I am going to explain how this works in more detail so maybe it helps assessing whether something is a bug or not.

When pressing a combination of keys that have a --release binding, scroll will wait until you release all of them before firing the --release binding. However, if at some point while holding some of those keys, you press another key that also triggers a binding (including the currently pressed keys), the original history is reset, and the current binding becomes active, as it should be,

I haven't found a problem with it yet, but if you find a problem or have a different proposal, please share. I would like to make this as robust as possible.

[larskotthoff]

This sounds good to me. I've experimented some more and I think I found the source of the issue -- when I do it very quickly, sometimes I release alt first and tab immediately after that. This means that the release binding for alt doesn't trigger (correctly). Fixed by adding another release binding inside the mode for alt+tab. That seems to be working as expected in all cases now.