Re: cross platform xinerama API proposal

["Matthias Clasen" <matthiasc poet de>]

----- Original Message -----
From: "Erwann Chenede" <Erwann Chenede Sun COM>
To: <gtk-devel-list gnome org>
Cc: <matthiasc poet de>
Sent: Thursday, November 22, 2001 4:49 PM
Subject: Re: cross platform xinerama API proposal

> Yes, it very similar, the main differences are :
> - the xinerama functions are logically attached to the GdkDisplay
>  instead of GdkScreen.

But that is not really future-save, isn't it ? While the current
implementation allows only one screen
per Xinerama-enabled display, it would be a very logical extension of the
Xinerama
extension to allow multiple screens, each stretching across multiple
monitors (IIRC, HP had
such a system long ago, called SLS - Single Logical Screen).

I think the functions make more sense on the Screen object. Unless we want
to restrict gdk
applications to a single screen for all times (I must admit that
applications which can make use
of multiple screens are *very* rare. Emacs can do it, I think, but that is
not really an argument,
I guess.)

> I'll incorporate your fix for the menu and window placement at the
> same time, if it is ok with you.

Great, go ahead.

>
>The only thing I would do differently is avoiding the special-case for
>non-Xinerama mode. Why not just have gdk_display_get_num_monitors and
>gdk_display_get_monitor_geometry return 1 and the screen geometry,
>respectively ?

Yes, this makes sense.

> gdk_display_use_virtual_screen is just a convenience function,
> I find more "natural" to have a test function instead of having
> to use gdk_display_get_num_monitors to check if the GdkDisplay
> is using xinerama like mode.

Makes sense, I guess, although the naming is debatable.

Thanks,

Erwann

[ I speak for myself, not for my employer. ]
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
  Erwann Chénedé, Sun Microsystems Ireland
  Desktop Applications & Middleware Group
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=