Re: Problems with ratpoison keys being passed to web applications in Firefox / Conkeror

Hi,

thanks for all the answers so far!

On Sat, Mar 13, 2010 at 10:12:22AM -0800, Shawn Betts wrote:
> > So I'd like to know: What goes wrong? Anyone here has an idea?
> 
> Is it a race condition? or does it happen even if you press the keys
> really slowly?

Even then, yes. Even tried 2 seconds gap between pressing Ctrl and t.

> Dunno why you'd get t and not C-t, though. Is your control key wearing
> out?

Not really. Would notice that quickly with ratpoison as primary window
manager on that box. :-)

But I guess, Bernhard can explain that:

On Sat, Mar 13, 2010 at 04:29:22PM +0100, Bernhard R. Link wrote:
> * Axel Beckert <[email protected]> [100313 15:16]:
> > I often switched to some xterm during the talk. And always when I
> > switched back to my slides with C-t C-t, the second C-t got passed as
> > "t" (not C-t) to the JavaScript running in the web browser. And since
> > "t" in S5 switches between slide show and web page view, I always have
> > to press "t" to get back to my slides.
> 
> I think there was a problem (and maybe still is), that ratpoison only
> catches the key-press, but not the key-release. (Perhaps it is still
> there, as I also do not know how that could be fixed).

That sounds very probable.

> This should only be a problem if some program reacts to a key-release
> without previously getting a key-press event.

Indeed. Will check what S5 does and if that's the case, file a bug
report against S5.

> Just some wild guessing...

I like that. ;-)

On Sat, Mar 13, 2010 at 06:03:26PM -0500, Mark Eichin wrote:
> although I don't see the bug (the only times I've seen a key after c-t
> leak through, it's a heavily loaded machine, and ratpoison never sees
> the key, which is what you'd expect...) I also use top-level bindings,
> and you might find them useful to get around it:

Yeah, my workaround is to use an interactive window list for now. But
working around that bug won't help to getting it fixed. ;-)
 
> definekey top M-Right next
> definekey top M-Left prev
>
> (which matches the linux-console vt-switch bindings.)  Since that's a
> single keystroke, it might avoid the problem...

Possible, yes. But I really non-top-level bindings. M-Right and M-Left
is e.g. used in Emacs for forward-word and backward word, and in irssi
it switches windows (like on the Linux VT console, too :-).

		Regards, Axel
-- 
Axel Beckert - [email protected], [email protected] - http://noone.org/abe/