WINDOWPATH environment variable?

[Samuel Thibault]

Samuel Thibault, le Mon 05 Dec 2005 11:53:31 +0100, a écrit :
> Carsten Haitzler (The Rasterman) wrote:
> > now if both are actually focused - how do u know which should have
> > brail terminal write access? the windowpath doesnt solve this
> 
> Of course it _does_ (I tested this case many times), because it is
> recursive.

And note that this sort of problems _already_ arose with just one X
server: applications may be running on VT1, VT2, etc, and on one window
of the X server running on VT7.

Applications running on VTs have an empty WINDOWPATH, and can determine
the VT number they're running on by themselves.  Applications running
on the X server have WINDOWPATH=7, and can determine the window they're
running in by themselves.  In addition to that, we run a little
"xbrlapi" program that tracks the X focus (with WINDOWPATH=7 too).

All of them give information to a centralized braille server that can
easily select the correct "current" application by querying the kernel
for the current VT number, and then, if it happens to be 7, use focus
information from the xbrlapi program.

There is no need for an X extension here, and it of course works
recursively with recursive Xnests. Even the screen case is covered.

That won't solve the "detaching" capability of screen/VNC, but the more
general (and difficult) solution that I proposed will solve this on the
long run.

Regards,
Samuel