Le lundi 07 avril 2008 à 22:01 +0200, Nicolas Mailhot a écrit : > Le lundi 07 avril 2008 à 21:39 +0200, Ralf Ertzinger a écrit : > > Hi. > > > > On Mon, 07 Apr 2008 13:58:57 +0200, Denis Leroy wrote > > > > > I think you summarized it pretty well. To be accurate, the VMWare > > > problem is not DPI reporting (which is correct, as per xrdb -query), > > > but the screen *size* reporting (in millimeters). > > > > Well... what size does a virtual screen have? > > vmware should probably match the virtual screen pixel size to the > current display real pixel size, and report the corresponding physical > size. Since it matches a virtual pixel to a real pixel, that would be > the correct thing to do IMHO. To elaborate a little more, I think you have close-to-eyes screens and distant screens. The dpi of anything in the first category should never be manipulated and kept to the real value (as in, take a ruler, measure physical size, count pixels, make ratio). That takes care of computer screens and embedded gadgets. The dpi of anything is the second category can not easily be measured. Computing an ideal perceived dpi value would need something to measure distance to viewers, assuming the distance is not dynamic and every viewer is at the same distance. So "angles of view" are not a realistic engineering concept. Instead, you can get by with heuristics, like anything in SD video resolution is ~ x dpi equivalent, anything in HD resolution is ~ y dpi equivalent, projectors that use a computer resolution are ~ 100 dpi equivalent (for now, that will probably change in the future), etc. And for virtual screens you match to whatever pixel density the actual screen you're rendering on uses. And all this should be done at the kernel/Xorg level, with app writers focusing on a way to share font size prefs (via xsettings, dbus or whatever is the cool tech of the day), instead of second-guessing the system (badly). -- Nicolas Mailhot
Attachment:
signature.asc
Description: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e?=
-- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
