[PATCH randrproto] Add "Border" and "BorderDimensions" properties

[Aaron Plattner]

On Tue, Aug 02, 2011 at 01:54:39PM -0700, Stéphane Marchesin wrote:
> 2011/8/2 Aaron Plattner <aplattner at nvidia.com>:
> > On Tue, Aug 02, 2011 at 01:15:19PM -0700, Stéphane Marchesin wrote:
> >> On Fri, Jul 29, 2011 at 12:27, Aaron Plattner <aplattner at nvidia.com> wrote:
> >> > Signed-off-by: Aaron Plattner <aplattner at nvidia.com>
> >> > ---
> >> > Does this interface make sense? I tried to describe what it should do as
> >> > accurately as possible.
> >> >
> >> > I'm working on a corresponding change to the server to make it read this
> >> > property and apply it during modesets. This is option #3 as described in
> >> > http://lists.x.org/archives/xorg-devel/2011-July/023888.html
> >> >
> >> >  randr.h        |    2 +
> >> >  randrproto.txt |   87 ++++++++++++++++++++++++++++++++++++++++++++++++++++---
> >> >  2 files changed, 84 insertions(+), 5 deletions(-)
> >> >
> >> > diff --git a/randr.h b/randr.h
> >> > index 41aedd5..9cb9403 100644
> >> > --- a/randr.h
> >> > +++ b/randr.h
> >> > @@ -181,5 +181,7 @@ typedef unsigned long       XRandrModeFlags;
> >> >  #define RR_PROPERTY_CONNECTOR_NUMBER   "ConnectorNumber"
> >> >  #define RR_PROPERTY_COMPATIBILITY_LIST "CompatibilityList"
> >> >  #define RR_PROPERTY_CLONE_LIST         "CloneList"
> >> > +#define RR_PROPERTY_BORDER             "Border"
> >> > +#define RR_PROPERTY_BORDER_DIMENSIONS  "BorderDimensions"
> >> >
> >> >  #endif /* _RANDR_H_ */
> >> > diff --git a/randrproto.txt b/randrproto.txt
> >> > index ce462ea..03a5351 100644
> >> > --- a/randrproto.txt
> >> > +++ b/randrproto.txt
> >> > @@ -138,6 +138,12 @@ underlying hardware to clients
> >> >      providing the server a complete configuration for appropriate
> >> >      resource management.
> >> >
> >> > +   • An optional Border property. This property allows a client to
> >> > +     specify that the viewport of the CRTC is smaller than the actve
> >>
> >> active
> >
> > Whoops, thanks.
> >
> >> > +     display region described its mode.  This is useful, for example,
> >> > +     for compensating for the overscan behavior of certain
> >> > +     televisions.
> >> > +
> >> >  The first two additions, per-crtc pixmaps and sprite transforms are
> >> >  designed to solve two problems:
> >> >
> >> > @@ -1006,8 +1012,8 @@ dynamic changes in the display environment.
> >> >
> >> >        'width' and 'height' indicate the size of the area within the screen
> >> >        presented by this CRTC. This may be different than the size of the
> >> > -       mode due to rotation. They will be set to 0 when the CRTC
> >> > -       is disabled.
> >> > +       mode due to rotation, the projective transform, and the Border property
> >> > +       described below.  They will be set to 0 when the CRTC is disabled.
> >> >
> >> >        'mode' indicates which mode is active, or None indicating that the
> >> >        CRTC has been disabled and is not displaying the screen contents.
> >> > @@ -1585,8 +1591,8 @@ factors, such as re-cabling a monitor, etc.
> >> >        rotation: ROTATION;             new rotation
> >> >        x: INT16                        x position of CRTC within screen
> >> >        y: INT16                        y position of CRTC within screen
> >> > -       width: CARD16                   width of new mode
> >> > -       height: CARD16                  height of new mode
> >> > +       width: CARD16                   width of new configuration
> >> > +       height: CARD16                  height of new configuration
> >> >  └───
> >> >        This event is generated whenever the CRTC configuration is changed
> >> >        and sent to requesting clients. 'timestamp' indicates when the
> >> > @@ -1594,7 +1600,9 @@ factors, such as re-cabling a monitor, etc.
> >> >        event. 'mode' is the new mode, or None if the crtc is disabled.
> >> >        'x' and 'y' mark the location in the screen where this CRTC
> >> >        is reading data. 'width' and 'height' indicate the size of the
> >> > -       mode. 'x', 'y, 'width' and 'height' are all zero when 'mode' is None.
> >> > +       CRTC viewport, which is the mode size adjusted by the optional
> >> > +       Border output property described below. 'x', 'y, 'width' and
> >> > +       'height' are all zero when 'mode' is None.
> >> >
> >> >        This event is sent whenever the monitor's configuration changes
> >> >        or if a new monitor configuration becomes available that was
> >> > @@ -1766,6 +1774,68 @@ doesn't handle a mandatory property correctly.
> >> >        a signal change (e.g. TV formats).  Clients are allowed to change the
> >> >        properties in order to select a different signal subformat.
> >> >
> >> > +    "Border"                   aka RR_PROPERTY_BORDER
> >> > +       Type:                   int16 [n]
> >
> > I'm thinking maybe I should change this to uint16, to match the type of the
> > width/height fields of MODEINFO.  I don't know why you'd want a 65535x65535 mode
> > with a 1x1 active area, but it seems silly to disallow it in the protocol.
> >
> 
> I thought about it, but didn't see any unsigned int anywhere else in
> the file. Not sure if that was intentional.

You mean in properties?  It's definitely possible, using type XA_CARDINAL.
It might be worth changing these types to the corresponding type atoms &
formats just for clarity's sake.

The mode sizes are CARD16, so they'd corresponding to property type
XA_CARDINAL, format=16.

> >> > +       Flags:                  Immutable
> >>
> >> Hmm, why Immutable?
> >
> > Immutable just prevents clients from reconfiguring or deleting the
> > property.  It doesn't prevent clients changing the size or contents of the
> > property (I know, that confused me too).
> >
> 
> Ah ok.
> 
> Stéphane