EDID decoding in libXrandr?

[Soeren Sandmann]

Keith Packard <keithp at keithp.com> writes:

> On Thu, 2007-11-15 at 18:59 +0100, Soeren Sandmann wrote:
> 
> > The monitor name displayed to the user should be computed based on
> > both the display descriptor strings, and on the vendor ID, depending
> > on what information is available. 
> 
> All I was suggesting was that the server should probably provide much of
> what you want in a EDID-indepenent fashion, so that the same tools will
> work with non-EDID monitors.

Alright, that makes sense. 

> If you've got a list of identifying information that you'll be pulling
> out of EDID, we can see which make sense to have as output
> properties.

For now, 

- the vendor id (DEL, SNY etc)

- the physical size of the screen (I believe this information is
  already provided in other ways, but for some reason it doesn't
  always agree with the EDID information

- the display descriptor strings

> > I also need the ability to compute an identifier that will allow the
> > client to recognize whether a monitor we have seen before, or a
> > sufficiently similar one, is plugged in so that the mode can be set
> > with no user interaction required at all.
> > 
> > Why should the server be involved in these policy questions? Just give
> > me the EDID information; parsing it is hardly rocket science.
> 
> RandR 1.3 will have a fairly long list of 'standard' properties to try
> and make user interfaces more consistent across different video drivers
> and monitors; if you've got information you're interested in, having us
> standardize the content and format will help not only your application,
> but video driver developers as well.

I don't know how to compute these identifiers yet. I suspect the
algorithm will have to be tweaked as we learn more about what monitors
and projectors are out there. 



Soren