Bug 16979

I had a closer look into this.

As far as I understand, xkbcomp cannot compile files in "rules/".
Sergey, is that indeed the case?

Sam, when you run 

/opt/xserver-xir-install/bin/xkbcomp -w 10 -a -llpRh /opt/xserver-xir-install/share/X11/xkb/\*

do you get any output? Return code should be 0.

My understanding at the moment is that xkbcomp should not be invoked on "rules/base". Whichever script tries to invoke xkbcomp with "rules/base", is the one at fault.

(In reply to comment #1)
> I had a closer look into this.
> 
> As far as I understand, xkbcomp cannot compile files in "rules/".
> Sergey, is that indeed the case?
> 
> Sam, when you run 
> 
> /opt/xserver-xir-install/bin/xkbcomp -w 10 -a -llpRh
> /opt/xserver-xir-install/share/X11/xkb/\*
> 
> do you get any output? Return code should be 0.
> 
> My understanding at the moment is that xkbcomp should not be invoked on
> "rules/base". Whichever script tries to invoke xkbcomp with "rules/base", is
> the one at fault.
> 

Yes I do, here is a sample:

-d------ -------- /opt/xserver-xir-install/share/X11/xkb/geometry/winbook
-d------ -------- /opt/xserver-xir-install/share/X11/xkb/types/basic
-dp----- -------- /opt/xserver-xir-install/share/X11/xkb/types/numpad
--p----- -------- /opt/xserver-xir-install/share/X11/xkb/types/numpad(mac)
--p----- -------- /opt/xserver-xir-install/share/X11/xkb/types/numpad(microsoft)
-dp----- -------- /opt/xserver-xir-install/share/X11/xkb/types/mousekeys
-dp----- -------- /opt/xserver-xir-install/share/X11/xkb/types/level5
-d------ -------- /opt/xserver-xir-install/share/X11/xkb/types/complete
-dp----- -------- /opt/xserver-xir-install/share/X11/xkb/types/iso9995
-dp----- -------- /opt/xserver-xir-install/share/X11/xkb/types/extra
--p----- -------- /opt/xserver-xir-install/share/X11/xkb/types/extra(keypad)
--p----- -------- /opt/xserver-xir-install/share/X11/xkb/types/extra(mac_keypad)
-dp----- -------- /opt/xserver-xir-install/share/X11/xkb/types/cancel
-d------ -------- /opt/xserver-xir-install/share/X11/xkb/types/default
-dp----- -------- /opt/xserver-xir-install/share/X11/xkb/types/pc
--p----- -------- /opt/xserver-xir-install/share/X11/xkb/types/caps(internal)
--p----- -------- /opt/xserver-xir-install/share/X11/xkb/types/caps(internal_nocancel)
--p----- -------- /opt/xserver-xir-install/share/X11/xkb/types/caps(shift)
--p----- -------- /opt/xserver-xir-install/share/X11/xkb/types/caps(shift_nocancel)


That looks like it is correct output. Perhaps xkbcomp should have a warning system when it is invoked on a particular file? I'll restart X to check if it works.

rules/base is not a valid file for xkbcomp. People are expected to read some docs or use GUI tools (like gnome-keyboard-properties).

> Perhaps xkbcomp should have a warning system when it is invoked on a particular file?

How? Just by the name? Or path? I'd say it would be very stupid

(In reply to comment #2)
> (In reply to comment #1)
> > I had a closer look into this.
> > 
...
> 
> That looks like it is correct output. Perhaps xkbcomp should have a warning
> system when it is invoked on a particular file? I'll restart X to check if it
> works.
> 

xkbcomp bails out in the flex/bison generated code when parsing "rules/base".
It would require some effort if we were to change the code there.

(In reply to comment #4)
> (In reply to comment #2)
> > (In reply to comment #1)
> > > I had a closer look into this.
> > > 
> ...
> > 
> > That looks like it is correct output. Perhaps xkbcomp should have a warning
> > system when it is invoked on a particular file? I'll restart X to check if it
> > works.
> > 
> 
> xkbcomp bails out in the flex/bison generated code when parsing "rules/base".
> It would require some effort if we were to change the code there.
> 

Clearly I'm misinformed then.

Regardless, your command for the xkb compiler worked and I thank you for telling me it. My VT switching is back, but some more complex things like the super key and volume keys are not at the moment. I'll revert xkbcomp and xkeyboard-rules to the most recent version and try again.

Thanks for your help. This probably isn't a bug then, just me being naive.

(Btw, 'rtfm' or that written in other words is not a helpful response, especially when the documentation I could find is convoluted and outdated ;-) )