W3C home > Mailing lists > Public > www-style@w3.org > July 2010

Re: [CSS21] Issue 149 - px vs. pt

From: David Singer <singer@apple.com>
Date: Wed, 7 Jul 2010 16:35:12 -0700
Message-Id: <D0489D93-044C-45FF-A09F-BAC2995E7A5A@apple.com>
To: "www-style@w3.org list" <www-style@w3.org>
OK, can I summarize the state of play?  I hope I get it right, because I feel honor-bound to help close this issue.  Choices seem to be:

1) Elika's write-up: http://fantasai.inkedblade.net/style/specs/css2.1/px-unit
Basically, puts all units in their normal relative values, except that resolution is 'deemed' to be 96px/in always.  In addition, devices can choose whether inches-match-inches at the viewing surface (anchor method 1) or whether the CSS pixel approximates the viewing angle subtended by a 96 dpi pixel at arms' length.  Devices might choose one or both of
a) presenting the user with a viewing surface which is larger than the physical surface, and requires panning around in the UI
b) presenting the user with a pre-scaled view, where the reference pixel is about right but length units do not match physical lengths on the display surface (e.g. scaling pages down for small, high-res, closely-viewed handheld devices, and scaling pages up for light-array ballpark billboards and other low-res remotely-viewed devices).

2) Add to that a 'resolution' metric, which has the fixed value 96.  Then add a whole set of metrics; for the sake of this exposition, take the existing ones and put 'true' in front of them (truepx, truein, truecm etc.).  These *differ* from (1) in that 
a) if there is a well-defined viewing surface, no matter where or what it is, real physical inches match inches on that surface, and so on;
b) trueresolution reports the truth as well, as device pixels per true inch (we might need fractional or real values here).
c) the actual size of the physical viewing surface is reported in width and height, not the size of some logical surface over which the user can pan.

If the device does not sensibly have a display surface, then trueXX could be the same as XX, but the display surface must still fit on the physical surface.

3) Try to reverse the clock and get back to the point where resolution IS a variable, and so on.  This would result in something in between (1) and (2), I suspect, with only one set of measurements.  I don't have, and have not seen, a fully worked proposal for this.

Are there other choices?

David Singer
Multimedia and Software Standards, Apple Inc.
Received on Wednesday, 7 July 2010 23:35:45 UTC

This archive was generated by hypermail 2.3.1 : Monday, 2 May 2016 14:38:37 UTC