- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Sat, 14 Aug 2010 11:15:41 -0700
- To: Brad Kemper <brad.kemper@gmail.com>
- Cc: Doug Schepers <schepers@w3.org>, "Dr. Olaf Hoffmann" <Dr.O.Hoffmann@gmx.de>, public-fx@w3.org, www-style@w3.org
On Sat, Aug 14, 2010 at 10:51 AM, Brad Kemper <brad.kemper@gmail.com> wrote: > On Aug 14, 2010, at 9:56 AM, Doug Schepers wrote: >> Wouldn't it be possible to meet both use-cases by adding a property, something like 'unit-space':'realteive*|absolute', where the default is to use these abstracted units (for necessary bugwards compatibility), and the 'absolute' option would do its best to render the physical size as indicated? This would be something like dealing with transforms or viewboxes. > > That seems dangerous. An author might routinely put 'unit-space:absolute' in his style sheet because he does fully understand it, and it seems reasonable and looks OK on his 280 ppi monitor. Then the page is viewed by someone else on a 72 ppi monitor, and all the 1px borders become sub-pixel widths, and all his 15px square icons look like crap, and everything renders slower. The Web is supposed to be fault tolerant, and this would seem to instead invite unintended rendering. To be fair, I think the idea would be that unit-space:absolute would break the connection between px and physical units. So changing the dpi of the screen wouldn't affect the size of things specified in px. (My point about it not solving a real problem still stands.) ~TJ
Received on Saturday, 14 August 2010 18:16:35 UTC