Resolution indpendence

Steven J Newbury steve at snewbury.org.uk
Fri Jun 27 07:20:45 PDT 2008


On Fri, 2008-06-27 at 15:36 +0200, Nicolas Mailhot wrote:
> Le Ven 27 juin 2008 15:13, Glynn Clements a écrit :
> >
> >
> > Nicolas Mailhot wrote:
> >
> >> >> The upcoming GNOME will simply set it to 96.
> >> >
> >> > SRSLY?  That would be a regression.  Right now GNOME nicely
> >> detects my
> >> > 114dpi screen and uses right size fonts.  96 would look really
> >> small.
> >>
> >> +1
> >>
> >> 1. Exposing correct DPI is hard
> >
> > Or impossible, if different portions of the "logical" screen have
> > different resolutions (as can happen with multiple monitors),
> 
> So you expose the characteristics of each portion. I believe the
> colour corrections guys want the same thing for colour information.
> It's useless to pretend an uniform plane when the hardware is in fact
> different.
> 
> > or if
> > you simply cannot determine the physical dimensions of the screen
> > (missing or incorrect EDID,
> 
> That's a bug. We don't ignore features because bugs exist.
> 
> > "virtual" screens (VNC etc)
> 
> You take the characteristics of the actual hardware the virtual screen
> is projected on
> 
> > projectors).
> 
> Those need TV mode and should not be treated as actual computer
> screens in the DPI sense. And you never actually do a screen wall
> composed both of actual screens and projections, so treating them
> differently is not problem.
> 
> >
> >> but is necessary for correct text rendering,
> >
> > I take issue with that. I know that a lot of people are emotionally
> > invested in this being true,
> 
> I think Behdad invested far more than emotions in text rendering those
> past years, and in fact I'm pretty sure that his investment most the
> apps you use would look different (and worse) today.
> 
> At my small level I do think I've made more than an emotional
> investment too.
> 
> We'd like the X guys to help our efforts not hinder them by going back
> to everything-is-96-dpi-even-if-it-isn't-really lala land
> 
> > but I've been forcing X to 75dpi from the
> > first day that it attempted to use the physical resolution, and it has
> > never caused me any problems
> 
> And people use 80x25 dumb terminals and are perfectly happy. Let's
> reduce X do 80x25 too. It works well.
> 
> > (whereas using the physical resolution certainly has caused problems).
> 
> Only because there are bugs, in particular deep GNOME bugs, that were
> hidden (and festered) by forcing 96 dpi some years ago (because if
> windows is doing it, it can't be wrong - oh, wait vista is not doing
> it anymore, so it was a stupid hack after all?)
> 
> >
> > Moreover, ...
> >
> > Ultimately, ...
> 
> That's all excuses to let the problems fester and hope they'll blow up
> on someone else's watch.
> 
> Yes there are problems. Yes it will be a painful transition. Doing it
> later won't make it any easier, though. It will only let more time for
> app authors to write code with broken pixel assumptions.
> 
 
I concur with everything Nicolas said.  Anybody who has ever done any
"Desktop Publishing" or half serious graphics work knows the importance of
having an accurate representation.  Just because GNOME UI
designers/coders can't be bothered to get it right really isn't a good
reason to break the desktops of people who use and *need* this feature
to work correctly.

Obviously randr-1.2 does make this much more difficult to achieve
although in a compositing environment certainly far from impossible.  It
may be necessary to limit "correct" operation to the zaphod and
composited cases, but this would probably satisfy most users avoiding
the hard case.




More information about the xorg mailing list