- From: Florian Rivoal <florian@rivoal.net>
- Date: Tue, 02 Apr 2013 14:01:10 +0200
- To: www-style@w3.org
On Thu, 21 Mar 2013 11:07:09 +0100, François REMY <francois.remy.dev@outlook.com> wrote: > Hi, > > > I was wondering what user agents were expected to do when a single page > is being displayed on two (or more) output devices at the same time. > [...] > > Any thought on that? I think it is an interesting question, made tricky by the fact that there is quite a bit of diversity hiding behind the 2 screen situation. One approach is simple duplication. Another one is merging the screens for a bigger surface. Yet another is what was done on the Nintendo DSi: one screen has the normal zoomed in view, and the other one has the zoomed out overview. One thing that all these situations have in common though is that in the end, we only ever have a single rectangular viewport. It may be duplicated (with or without the duplicate being zoomed differently), it may be displayed in separate chunks on several screens, but it is still a single rectangle whose dimensions in CSS pixels are unambiguously known. On the other hand, while the dimensions are not too tricky, the other characteristics may not be uniform. Color depth, monochrome or not, or resolution may differ across screens. Given the type of displays that are common these days, resolution is the key issue. What do you do if you are duplicating, and one screen has 1 physical pixel per css pixel, while the other one has 2? What about if you are stretching a single viewport over these two screens of different resolution? I don't really have an answer, and I don't think the existing specs point one way or the other. But as "retina" laptops get more common, we will increasingly run into situations where a computer is connected to 2 screens of different density, and we will have to find an answer. If we want to move away from this model of a single rectangular viewport, there is going to be a lot of difficulties. It might be possible to have essentially 2 viewports with completely different characteristics, providing 2 views onto the same document, but making that work would require significant work on the object model. I don't see any theoretical impossibility in having multiple views of the same document living in the same script environment, but this probably needs a lot more work than anybody is interested in putting in. - Florian
Received on Tuesday, 2 April 2013 12:01:34 UTC