Re: Keybindings execute twice?

[mmc maruska dyndns org (Michal Maruška)]

Johan Ingvast <ingvast md kth se> writes:

> Thanks for taking your time.  I had the program compiled and run, first on
> my Rh 7.3 installation (where sawfish works perfecly) and then on my Rh 9.
> The results came out with different behaviour for the v key.  
> The result from Rh 9 while pressing u-v-x-v-x-v:
>
> event received u release 1004596

> event received v press 1007085
> event received v release 1007593
> event received v press 1007166
> event received v release 1007166
> event received v press 1007593
> event received v press 1007593

yes, this 'storm' of repeated events is how the bug reveals itself.

For RH users see (the end of)
  https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=76959 

they have a link to updated (ERRATA) packages.

> I should also add something that might help in the diagnostics.  It is not
> only the M-C-t keybinding that gives the fault. The same thing happens if
> I bind it to M-C-a.  Also if I bind a key to some other command like 
> "Help about" the same thing happens: it gives me two help-about windows.

yes. The bug simply repeats the last (non modifier) key, while the keyboard (of
the X server) is 'grabbed'.  And the keybindings of sawfish are made via
'passive' grabs, i.e. when you press the key combination, a grab starts. If the
grab lasts less then the auto-key-repeating delay, nothing wrong happens,
otherwise ....


> Also: when I hit the keybinding it sometimes takes long time (several
> seconds) before anything happens, and sometimes nothing happens.

that's b/c you start the new processes.



if you don't have a possibility to install a new package, a quick fix is
_probably_ to either disable XKB in the X configuration, or (i'm not sure about
this) not change the auto-repeating rate, once the X is started.