MediaQueryEvaluator should use default style for evaluating "rem".

MediaQueryEvaluator should use default style for document for "rem".

Since the 'rem' unit is based on the initial font size when used within Media Queries (c.f.http://dev.w3.org/csswg/mediaqueries/#units),
we have to use StyleResolver::styleForDocument to obtain default style for document.

TEST=fast/css/media-query-basing-rem-unit.html
BUG=258200

[eseidel, 2013-07-11 05:37:59]

I don't feel particularly well informed about media queries, you may want to ask
Tab for his thoughts on the spec-compliance of this change, but the code lgtm!

You may also wish to consider testing what "rem" should do in the seamless case.

https://codereview.chromium.org/18360007/diff/1/Source/core/css/MediaQueryEva...
File Source/core/css/MediaQueryEvaluator.cpp (right):

https://codereview.chromium.org/18360007/diff/1/Source/core/css/MediaQueryEva...
Source/core/css/MediaQueryEvaluator.cpp:352: RefPtr<RenderStyle> defaultStyle =
StyleResolver::styleForDocument(document);
Is it OK that if this document is seamless it will be inheriting styles from
it's container in the document above it?

[kenneth.r.christiansen, 2013-07-11 08:36:35]

lgtm, but I also don't know about the seamless case

[TabAtkins, 2013-07-11 21:13:40]

On 2013/07/11 05:37:59, eseidel wrote:
> I don't feel particularly well informed about media queries, you may want to
ask
> Tab for his thoughts on the spec-compliance of this change, but the code lgtm!

Yes, this is the correct behavior.  (I filed a bug on this exact thing earlier
this week.)

> You may also wish to consider testing what "rem" should do in the seamless
case.
> 
>
https://codereview.chromium.org/18360007/diff/1/Source/core/css/MediaQueryEva...
> File Source/core/css/MediaQueryEvaluator.cpp (right):
> 
>
https://codereview.chromium.org/18360007/diff/1/Source/core/css/MediaQueryEva...
> Source/core/css/MediaQueryEvaluator.cpp:352: RefPtr<RenderStyle> defaultStyle
=
> StyleResolver::styleForDocument(document);
> Is it OK that if this document is seamless it will be inheriting styles from
> it's container in the document above it?

Interesting question!  I've asked www-style
<http://lists.w3.org/Archives/Public/www-style/2013Jul/0271.html>.



Separately, I'm not sure if window.resizeTo() is problematic or not.  If it's
fine, then don't worry about it, but if it would be faster/easier to avoid the
use of it, we can rewrite the test to instead set the font-size to something
*enormous*, like 100000px.  Then just do a MQ like (max-width: 1em), and if it
applies, the test fails.

[tasak, 2013-07-16 10:33:05]

Thank you for reviewing.

https://codereview.chromium.org/18360007/diff/1/Source/core/css/MediaQueryEva...
File Source/core/css/MediaQueryEvaluator.cpp (right):

https://codereview.chromium.org/18360007/diff/1/Source/core/css/MediaQueryEva...
Source/core/css/MediaQueryEvaluator.cpp:352: RefPtr<RenderStyle> defaultStyle =
StyleResolver::styleForDocument(document);
On 2013/07/11 05:38:00, eseidel wrote:
> Is it OK that if this document is seamless it will be inheriting styles from
> it's container in the document above it?

I added one more layout test for seamless iframe.
The test looks ok?