[OpenICC] Oyranos - proof of concept

[Jim Gettys]

On Fri, 2005-04-22 at 17:18 +0200, Kai-Uwe Behrmann wrote:
> Am 22.04.05, 08:31 -0400 schrieb Jim Gettys:
> 
> > On Fri, 2005-04-22 at 13:19 +0200, Kai-Uwe Behrmann wrote:
> > 
> > > A small utility - oyranos-gamma makes a call to the separate 
> > > liboyranos-moni.so lib to obtain, by a (poor) heuristic, the best matching 
> > > monitor profile name.
> > > Xcalib is called to upload the eventually contained vcgt tag to the 
> > > graphics card. The function to set the monitor profile is currently 
> > > utilised in icc_examin only. Oyranos-gamma can be automatically called 
> > > from a X init script like '.xinitrc'.
> > > 
> > > 
> > 
> > 
> > We'll need to ensure that hot-plug of monitors  on the fly is
> > possible/convenient. This is a very common operation these days given
> > laptops.  It is certainly something we plan to do for X in general.
> > 
> > D-Bus is the obvious candidate for part of the solution, given it's use
> > in the Linux hotplug subsystem.
> 
> The current state of Oyranos is proof of concept.
> 
> This means first searching answeres to some very basic questions :
> o is the APIs what we intent (paths, default profiles ...)?
>   flexible , fitting in others concepts, already too complex 
> o is xml currently an issue?
> o are there visible design flaws
> 
> 
> When Oyranos can get basic agreement here, then comes the next stage to 
> create an alpha version. At this stage I'd like to touch D-Bus as well. It 
> is to early to think too much about extending now.
> 
> To X:
> I think we can extend to realtime colour configuration for X . It 
> is ok to connect to D-Bus through a X dedicated library - liboyranos-moni.
> How can X / Oyranos transport the monitor profile information 
> networktransparently  (Craigs question?) Similiar to DDC?
> 

Arguably, unless the monitor profile information is huge, I'd recommend
making it a property on the root window of each screen, that
applications can retrieve.  What is more, applications can get
notification of changes to window properties, which would enable this to
be handled pretty easily in a cross toolkit fashion.  I think this is
roughly what was done in the past in the ill fated attempts at building
a CMS into Xlib.  And it means you can get started without any X server
infrastructure changes; your program can load such properties on its
own.

D-bus comes into play in telling the X server what is going on; after
all, changing a monitor has other repercussions that the X server may
need to know about directly.
					- Jim