Three recent blog postings on DRI3000

[Alexander E. Patrakov]

2013/8/4 Keith Packard <keithp at keithp.com>:
>
>> From the protocol viewpoint - OK. However, I think that at some point
>> some configuration request (similar to setting the xrandr primary
>> output? or even reusing that information, if it works) should be added
>> in order to make the "X server picks" case more predictable.
>
> Using the RandR primary output would be easy; although I'm not sure it
> would do exactly what you want in every case.
>
> I think the automatic selection heuristic could be a bit fancier and get
> closer more of the time though. I suspect what you really want is for
> the monitor containing the 'full screen' version of the window to be the
> synchronization source, which is to say the monitor which contains the
> greatest proportion of the window's pixels, and when multiple monitors
> show the entire window, then pick the monitor which has the fewest
> pixels from other windows visible.
>
> For multiple 'perfect clones', we could then fall back to the RandR
> primary output.
>
> Seems a bit complex; but is it less complex than adding yet another
> twisty configuration parameter?

I understand your concern about new configuration paramenets, and your
heuristic (using primary output only for disambiguation when
equally-sized monitors contain equal proportions of the window's
pixels) looks good enough. So let's use it.

-- 
Alexander E. Patrakov