Re: Add unredirect block / request hints

[Adel Gadllah <adel gadllah gmail com>]

Am 09.01.2013 18:29, schrieb Owen Taylor:
> On Sun, 2012-10-07 at 13:36 +0200, Adel Gadllah wrote:
> > Hi,
> >
> > I have attached an updated patch based on an email discussion with Martin.
> >
> >  From 047bae702be8180c50119660c858111c5088450f Mon Sep 17 00:00:00 2001
> > From: Adel Gadllah <adel gadllah gmail com>
> > Date: Tue, 14 Feb 2012 14:30:28 +0100
> > Subject: [PATCH] Add _NET_WM_BYPASS_COMPOSITOR and
> >   _NET_WM_DONT_BYPASS_COMPOSITOR
> >
> > Add properties that allows client windows to hint the compositor that they
> > should or should never be unredirected / run uncomposited.
> > ---
> >   wm-spec/wm-spec.xml | 35 +++++++++++++++++++++++++++++++++++
> >   1 Datei geändert, 35 Zeilen hinzugefügt(+)
> >
> > diff --git a/wm-spec/wm-spec.xml b/wm-spec/wm-spec.xml
> > index 87dc02d..858acbb 100644
> > --- a/wm-spec/wm-spec.xml
> > +++ b/wm-spec/wm-spec.xml
> > @@ -1578,6 +1578,38 @@ Rationale: This gives the compositing manager
> > more room for optimizations.
> >   For example, it can avoid drawing occluded portions behind the window.
> >           </para>
> >           </sect2>
> > +
> > +	<sect2><title>_NET_WM_BYPASS_COMPOSITOR</title>
> > +        <programlisting><![CDATA[
> > +_NET_WM_BYPASS_COMPOSITOR, ATOM
> > +]]></programlisting>
> > +        <para>
> > +The Client MAY set this property to hint the compositor that the window would
> > +benefit from running uncomposited (i.e not redirected offscreen).
> > +The compositing manager MAY ignore this hint.
> > +        </para>
> Why is this property an ATOM? The behavior of the patch you landed for
> Mutter is to treat it as a CARDINAL and ignore the property (treat it as
> unset) unless it has the value '1'.
>
> I think I would have replaced both properties with a single enumeration:
>
>   0 (or unset) - no preference
>   1 - please bypass
>   2 - please do not bypass
>
> which also would avoid the question of having both properties set.

This does indeed make sense. Will follow up with a new patch that uses an enumaration instead.

> If we
> do leave two properties (if anybody has already shipped it in this form
> or we think it makes more sense in the end), then I'd like the spec to
> explicitly say which one wins.
>
> I'd also like us to explicitly say that the compositing manager MAY
> bypass compositing for both fullscreen and non-fullscreen windows if
> bypassing is requested, but MUST NOT bypass if it would cause
> differences from the composited appearance.
OK.

Thanks,
Adel