RandR 1.3 additions?

[Alex Deucher]

On Jan 22, 2008 1:53 PM, Luc Verhaegen <libv at skynet.be> wrote:
>
> On Tue, Jan 22, 2008 at 01:32:23PM -0500, Alex Deucher wrote:
> > On Jan 22, 2008 1:26 PM, Luc Verhaegen <libv at skynet.be> wrote:
> > > On Tue, Jan 22, 2008 at 06:24:58PM +0100, Matthias Hopf wrote:
> > > > On Jan 22, 08 12:03:15 -0500, Alex Deucher wrote:
> > > > > > > could be used as the "standard" properties for choosing an encoder
> > > > > > > though.  That said I think in the long run having encoder objects is
> > > > > > > clearer.  It's be nice to know that the VGA port and the TV port
> > > > > > > shared the same actual encoder rather than just signal format
> > > > > > > "analog."
> > > > > >
> > > > > > Signal formats for VGA and TV differ significantly enough. "analog" is
> > > > > > certainly no valid description for a signal format, and hasn't been
> > > > > > proposed in the spec update.
> > > > >
> > > > > Right, but my point was that you want to know which encoder is used
> > > > > rather than just the signal format.  The encoder would dictate the
> > > > > signal format.    Although I suppose you'd want both since it would be
> > > > > nice to have a standard way of knowing the signal type as well.
> > > >
> > > > Hm, the encoder will vary from card to card. Yes, it would be preferable
> > > > if even the encoder was abstracted, though this would be very card
> > > > dependent. Feel free to add something related to my proposal, I don't
> > > > have a good feeling ATM how this should look like.
> > > >
> > > > Matthias
> > >
> > > "Encoder abstraction"?
> > >
> > > Hasn't "encoder" been "output" all along? Output is encoder, output is
> > > not connector.
> >
> > By outputs I mean randr 1.2 outputs.  It depends on the driver.  On
> > Intel they are one and the same as I don't think they support any
> > mixed encoder outputs.  radeonhd exposes encoders.  radeon exposes
> > connectors.  I'm not sure about mga.
> >
> > Alex
>
> The way i have been using the word output signifies encoder, and i have
> been using this naming excessively ever since early 2004. This is also
> what Dave Airlie cut out of the unichrome driver to fit only external
> i2c output devices, or encoders, and this is how these terms eventually
> entered the intel driver.
>

Believe it or not lots of people (including myself) have been thinking
about multi-head abstraction for quite a while, often independently.
It's awful presumptuous to say you are the origin of the "output"
concept.

> The fact that you saw no other option but to abuse output to mean
> connector means that there are shortcomings in the model. But this is no
> reason to change the original and most widespread meaning.

Where do you get original?  Just because you think output meant
encoder doesn't mean that's the "standard" understanding.  A randr
output is vague at best.  It' just as reasonable to associate output
with connector as to associate output with encoder.  As it's used in
randr, "output" seems to fit connector better than encoder.  It more
or less signifies a monitor, something that plugs into a connector.

Alex