alttab behaves incorrectly when Control_R or Alt_R is used to initiate window switch

[myelsukov]

I think the bug is here: https://github.com/sagb/alttab/blob/master/src/alttab.c#L521

The if thinks that modifier key is released so it closes the UI before exposing it.
The window is getting switched though.

[sagb]

How can I reproduce? alttab -mk Control_R works fine for me.
alttab -mk Alt_R does not, because my window manager uses Alt_R for its purpose. Perhaps you have the problem for the same reason?

[myelsukov]

1. Start alttab -mk Control_L -w 1 -d 1 -s 1
2. Now Control_L+Tab works as expected: a window with tiles is shown and you can switch between windows.
3. Try Control_R+Tab. A window with tiles is not shown, but the focus is moved to the next window. If you keep pressing and releasing Tab without releasing Control_R or if you keep pressing and releasing Control_R+Tab the focus will be switched between the same two windows.

May be this is a feature? 😁
But I think you have a problem with the line I mentioned in the bug report.

[sagb]

I doubt it's alttab bug. The line you show is very core logic, I don't see any error here. Have you managed to change/fix the issue by changing this line?
What's your window manager?

[myelsukov]

The logic is flawed. The Alt/Mod1 modifier is triggered by the either L_Alt or R_Alt and it properly intercepted by the alttab.
Then in that line you check if the L_Alt is still pressed. Well, it is not pressed, because I pressed R_Alt so you erroneously decide that L_Alt was released and remove the tiles window.

[sagb]

Ah, I got it: you write about the comment "Alt released" in that line. It doesn't mean "Alt" literally, but any modifier set by -mk. Sorry for misleading comment.
Alttab works with various -mk, including Control_R and Alt_R.

If you want to try to resolve your issue, look at window manager key bindings, the question you ignored.

[myelsukov]

I ignored question about WM as irrelevant 😁
If you are interested, then it's JWM and I personally control all keyboard bindings. Trust me, the keyboard binding has nothing to do with this problem.
The problem is the following:

When you bind the activation hotkey, you bind the keyboard modifier (Ctrl, Mod1, etc) independent on the position of the modifier key: left or right.
So, if you start alttab as alttab -mk Control_L -w 1 -d 1 -s 1 you can trigger it with both Controls: Left and Right (try it if you don't trust me).
But the line https://github.com/sagb/alttab/blob/master/src/alttab.c#L521 tests for the specific control only. In our case it's Left Control.
So, if I trigger alttab with Control_R it gets triggered, the tiles window is displayed then it immediately disappears because that line ⬆️ thinks that Control_L key was released and removes the window. In fact, Control_L was never pressed.

[sagb]

You are right, thank you for pointing out to this. Should be fixed now.

[myelsukov]

I commented on the commit 😁