Added primary input devices setting to blink to allow media queries for hover/pointer

Added primary input devices setting to blink to  allow media queries for hover/pointer

Added a "primary input devices" setting in Blink to allow implementation of hover:1 and pointer:fine|none in addition to hover:0 and pointer:coarse, already implemented in http://wkb.ug/87403. 

This setting is a bit flag, allowing multiple devices to be specified. It indicates
to Blink the available input devices that we expect the user is interacting with.

This CL was started and first reviewed in http://wkb.ug/87806


BUG=136119

[Rick Byers, 2013-04-10 14:57:50]

Using bitflags to represent the primary pointers seems like a great trade-off to
me (still simple, but gives us the expressiveness we want).

The CL description should be more detailed and contain some of the justification
we've talked about on the WebKit review.  Maybe something like:
--
Add mouse support to pointer and hover media queries

Add a setting for the browser to indicate which pointer devices it supports as
'primary input devices'.  Use this setting to implement pointer:fine,
pointer:none, and hover:1 queries (in addition to pointer:coarse and hover:0
which were previously implemented in http://wkb.ug/XXX).

This CL was started and first reviewed in http://wkb.ug/XXX
--

There should probably be a comment somewhere in the code (eg. in the setting)
clarifying the meaning of "primary pointer devices" - i.e. "input devices that
are/were present and are expected to be used".

Also I realized we should probably see if we need to go through the blink web
platform feature process for this (it's kind of a grey area between a bug in an
existing feature, and a new web feature).  I'll post to blink-dev.  Please hold
of on putting this in the CQ until we have agreement there.

https://codereview.chromium.org/13818030/diff/1/Source/WebCore/platform/chrom...
File Source/WebCore/platform/chromium/PopupListBox.h (right):

https://codereview.chromium.org/13818030/diff/1/Source/WebCore/platform/chrom...
Source/WebCore/platform/chromium/PopupListBox.h:86: bool primaryInputIsTouch;
this again implies that there's only a single primary input.  I'd say instead
'primaryInputIncludesTouch' or even just 'usesTouchscreen'.

https://codereview.chromium.org/13818030/diff/1/Source/WebCore/testing/Intern...
File Source/WebCore/testing/InternalSettings.cpp (right):

https://codereview.chromium.org/13818030/diff/1/Source/WebCore/testing/Intern...
Source/WebCore/testing/InternalSettings.cpp:489: comma_position =
pointerDevice.find(',', ix);
Is there any other code here or in WTF somewhere for parsing a comma separated
list that you can easily share with?  This is pretty trivial, but still would be
nice not to have explicit parsing code here if there's already some in a common
place.

https://codereview.chromium.org/13818030/diff/1/Source/WebKit/chromium/public...
File Source/WebKit/chromium/public/WebSettings.h (right):

https://codereview.chromium.org/13818030/diff/1/Source/WebKit/chromium/public...
Source/WebKit/chromium/public/WebSettings.h:59: PointerDeviceNone = 4
Document that None is mutually exclusive with all the other flags (perhaps it
should be first).  Other flags can occur in any combination (except Unkown is of
course not a flag).

[bokan, 2013-04-10 16:03:29]

https://codereview.chromium.org/13818030/diff/1/Source/WebCore/platform/chrom...
File Source/WebCore/platform/chromium/PopupListBox.h (right):

https://codereview.chromium.org/13818030/diff/1/Source/WebCore/platform/chrom...
Source/WebCore/platform/chromium/PopupListBox.h:86: bool primaryInputIsTouch;
On 2013/04/10 14:57:51, Rick Byers wrote:
> this again implies that there's only a single primary input.  I'd say instead
> 'primaryInputIncludesTouch' or even just 'usesTouchscreen'.

Done, changed to primaryInputIncludesTouch

https://codereview.chromium.org/13818030/diff/1/Source/WebCore/testing/Intern...
File Source/WebCore/testing/InternalSettings.cpp (right):

https://codereview.chromium.org/13818030/diff/1/Source/WebCore/testing/Intern...
Source/WebCore/testing/InternalSettings.cpp:489: comma_position =
pointerDevice.find(',', ix);
On 2013/04/10 14:57:51, Rick Byers wrote:
> Is there any other code here or in WTF somewhere for parsing a comma separated
> list that you can easily share with?  This is pretty trivial, but still would
be
> nice not to have explicit parsing code here if there's already some in a
common
> place.

Done, replaced with WTFString::split

https://codereview.chromium.org/13818030/diff/1/Source/WebKit/chromium/public...
File Source/WebKit/chromium/public/WebSettings.h (right):

https://codereview.chromium.org/13818030/diff/1/Source/WebKit/chromium/public...
Source/WebKit/chromium/public/WebSettings.h:59: PointerDeviceNone = 4
On 2013/04/10 14:57:51, Rick Byers wrote:
> Document that None is mutually exclusive with all the other flags (perhaps it
> should be first).  Other flags can occur in any combination (except Unkown is
of
> course not a flag).

Done.