Support for multiple independent mouse cursors
Carsten Haitzler (The Rasterman)
raster at rasterman.com
Wed Apr 26 01:27:10 PDT 2006
On Wed, 26 Apr 2006 10:15:44 +0200 Samuel Thibault
<samuel.thibault at ens-lyon.org> babbled:
> Carsten Haitzler, le Wed 26 Apr 2006 17:18:55 +0900, a écrit :
> > On Wed, 26 Apr 2006 10:01:13 +0200 Samuel Thibault
> > <samuel.thibault at ens-lyon.org> babbled:
> >
> > > Carsten Haitzler, le Wed 26 Apr 2006 17:05:01 +0900, a écrit :
> > > > On Wed, 26 Apr 2006 09:37:53 +0200 Samuel Thibault
> > > > <samuel.thibault at ens-lyon.org> babbled:
> > > >
> > > > > Peter Hutterer, le Wed 26 Apr 2006 10:41:34 +0930, a écrit :
> > > > > > The only way I found around that is the following:
> > > > > > modify X lib so that each time a window is created it also sets an
> > > > > > atom on this window to the PID of the process that opened it.
> > > > >
> > > > > For systems that support it (linux at least), you can use SO_PEERCRED
> > > > > on unix sockets for getting pid, uid and gid of the peer process.
> > > >
> > > > not that that is of any use in X as the WM doesn't have a
> > > > fd /connection to the client. only x does. and wm only connects to x.
> > >
> > > Ok, but the X server can set the property itself and mark it as
> > > read-only. Then the WM can really trust the property.
> >
> > sure - it could, in theory, if x and the process run on the same box.
> > alternately you can do some nice LD_PRELOAD fun with xlib to force windows
> > to have such properties - and much much much more. :) no need to modify x
> > or even xlib.
>
> If you want trust, you need to have some mecanism in the X server.
but since it's the WM that needs the info... it needs to provide the mechanism.
maybe i'm just used to making a wm work on any xserver - irrespective of its
extensions and patches.
--
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler) raster at rasterman.com
裸好多
Tokyo, Japan (東京 日本)
More information about the xorg
mailing list