Re: [RP] Bug with rdesktop, and programs wanting Ctrl-T

[Casey Allen Shobe]

> These issues with key release events have caused problems in the past.
> It's a bit of a difficult one to fix. I'm sooort of inclined to call it
> a bug in XNest. Take this for an example:
> 
> You are in a standard WM that uses the mouse to switch windows. You are
> in XNest and you press CTRL, then click another window which shifts the
> focus, then release. The window that now has focus gets the ctrl release
> event not Xnest and now you have the same problem. While I haven't
> *actually* done this, I imagine this is what would
> happen. If it does happen, it seems to me that Xnest is buggy. What do
> you think?

Hmmmm, I'm not certain.  Do the key release problems only occur with Xnest?  
In that case, if the situation you describe is reproducible, then I'd agree 
that it's an Xnest problem.  But the question comes to mind - what exactly 
is Xnest doing wrong?  By only trapping the signals it's being sent, this 
seems like normal operation of an X server.  Similarly, if I'm in a regular 
X, hold down a key and somehow release it subverting X11, then X11 will 
likely think the key is still held down.  Key press/release timing is 
important (i.e. when you want to press Ctrl-T), so I think the problem 
itself may lie in ratpoison, if Xnest is behaving as a normal X11 server 
(which is it's intention).

What about this:  When ratpoison recieves a Ctrl-T, then it could send a 
keyrelease to the X11 server.  This would break lazy-ctrl keying, i.e. 
Ctrl-T,N, and would enforce Ctrl-T, Ctrl-N.  A possible alternative to 
preserve lazy-ctrl keying would be to immediately follow the keyrelease 
with a keypress signal, so that even if the user kept holding the button 
down, ratpoison would register it as a second CTRL keypress after the Ctrl-
T, while the application would register the keyrelease.  This could 
possible present a problem if the user lets go of the Control key 
IMMEDIATELY before the keyrelease, keypress signals could be sent, but I 
don't know how likely (or if at all possible) that is...

- Casey