Re: [screen-orient] why not provide at CSSOM API to CSS Device Adaptation instead?

There is a distinction to be made between DIPs and CSS pixels. Take a
portrait iPhone 4 for example:
- screen is 640 device pixels wide
- device pixel ratio is 2

CSS pixels are affected by pinch zooming:

If you use <meta name="viewport" content="width=device-width,
initial-scale=1"> then:
- screen.width is fixed at 320 DIPs
- document.documentElement.clientWidth is fixed* at 320 CSS pixels
- window.innerWidth is initially 320 CSS pixels, but decreases when the
user pinch zooms in, since zooming in makes CSS pixels larger, and hence
less of them fit onscreen at once.

CSS pixels are also affected by viewports:

If you have no meta viewport or @viewport rule, then mobile browsers will
set your page width to a fallback value of around 980 CSS pixels, and so:
- screen.width is still fixed at 320 DIPs
- document.documentElement.clientWidth is fixed* at 980 CSS pixels
- window.innerWidth is initially 980 CSS pixels, but decreases when the
user pinch zooms in. When zoomed to a scale of 1, CSS pixels are the same
size as DIPs, window.innerWidth is 320, and content is now as legible as it
would have been if width=device-width had been used.

*: the reason that document.documentElement.clientWidth remains fixed, even
though it's unit is CSS pixels, is that when you zoom you're no longer
measuring the same distance, instead the distance being measured (the page
width) scales in proportion to the page.


On Thu, Jun 6, 2013 at 10:04 AM, Kenneth Rohde Christiansen <
kenneth.christiansen@gmail.com> wrote:

> As I understand it, the CSS specs never refer to device pixels, but
> always to CSS pixels which are basically your DIPs, so the confusion
> only arises from the fact that browsers did DPI adjustment (upscaled
> the content to use real CSS values in CSS units - aka DIPs) but forgot
> to do it for everything.
>
> As Apple did the upscaling in their core system/toolkit, everything
> stayed in CSS units from day one, where as other browsers such as
> Android and Qt (Nokia N9) did the upscaling in the browser itself and
> forgot to adjust all web facing values.
>
> Kenneth
>
> On Wed, Jun 5, 2013 at 4:34 PM, John Mellor <johnme@chromium.org> wrote:
> > Sounds reasonable. I think the main things that are missing are
> explanations
> > of the knock-on effects of Device Adaptation (and the meta viewport tag).
> >
> > For example the CSSOM View module claims that all its dimensions are in
> CSS
> > pixels. But if you look at what mobile browsers return for something like
> > screen.width, they either return Device Independent Pixels (DIPs) or
> > physical device pixels. It's only for things like window.innerWidth that
> > mobile browsers actually return a value in CSS pixels. Now, returning
> DIPs
> > does in fact best match the intent of the spec for screen.width (indeed
> the
> > definition of a CSS pixel is actually the definition of a DIP, and they
> used
> > to be the same thing until pinch zoom and viewports made them scale
> > independently); but the CSS specs need to accept that there are more
> kinds
> > of pixels than there used to be, and fix these ambiguities, if we want
> > mobile browsers to converge on a single behaviour.
> >
> >
> > On Wed, Jun 5, 2013 at 2:59 PM, Kenneth Rohde Christiansen
> > <kenneth.christiansen@gmail.com> wrote:
> >>
> >> I think it should be. You think anything is missing?
> >>
> >> Kenneth
> >>
> >>
> >> On Wed, Jun 5, 2013 at 3:51 PM, Simon Pieters <simonp@opera.com> wrote:
> >>>
> >>> On Wed, 24 Apr 2013 13:00:48 +0200, Kenneth Rohde Christiansen
> >>> <kenneth.r.christiansen@intel.com> wrote:
> >>>
> >>>> I support adding some CSSOM API's for CSS Device Adaptation, but I
> would
> >>>> not do so for the viewport meta tag, which has its share of issues.
> >>>
> >>>
> >>> There's currently
> >>> http://dev.w3.org/csswg/css-device-adapt/#dom-interfaces
> >>>
> >>> Is that sufficient?
> >>>
> >>> --
> >>> Simon Pieters
> >>> Opera Software
> >>>
> >>
> >>
> >>
> >> --
> >> Kenneth Rohde Christiansen
> >> Senior Engineer, WebKit, Qt, EFL
> >> Phone  +45 4294 9458 / E-mail kenneth at webkit.org
> >>
> >> ﹆﹆﹆
> >
> >
>
>
>
> --
> Kenneth Rohde Christiansen
> Senior Engineer, WebKit, Qt, EFL
> Phone  +45 4294 9458 / E-mail kenneth at webkit.org
>
> ﹆﹆﹆
>