Add support for high-DPI hosts under Mac OS X.

Add support for high-DPI hosts under Mac OS X.

- Moves enumeration of display configuration(s) to helper classes, which fetch each display's logical and device-native resolutions.
- Updates VideoFrameCapturerMac and EventExecutorMac to use the helper.

This is a re-land of r178318, which was reverted because it didn't build under the Mac OS X 10.6 SDK.

BUG=135081
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=178761

[Wez, 2013-01-24 00:12:24]

PTAL

This now builds on 10.6 but I still need to find a 10.6 system to actually try
it out on.

[Jamie, 2013-01-24 00:20:18]

Changes LGTM.

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
File remoting/capturer/mac/desktop_configuration.mm (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.mm:13: MAC_OS_X_VERSION_MAX_ALLOWED
< MAC_OS_X_VERSION_10_7
Nit: indent the continuation line.

[Nico, 2013-01-24 00:26:58]

lgtm

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
File remoting/capturer/mac/desktop_configuration.h (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.h:28: // Bounds of this display in
logical (72dpi) coordinates.
We call "logical coordinates" "dips" or "device-independent pixels" in chromium
code (following android's lead), and most things are in dips by default and only
stuff in pixels is called out as "pixel_foo".

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.h:38: float logical_to_pixel_scale;
For what it's worth, we call this "device_scale" in most other places in
chromium and webkit code.

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.h:58: SkIPoint dpi;
I'm not sure if dpi is a useful metric on OS X, isn't it always 72?

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
File remoting/capturer/mac/desktop_configuration.mm (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.mm:24: const int kDefaultDPI = 72;
nit: no indent in namespaces

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/v...
File remoting/capturer/video_frame_capturer_mac.mm (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/v...
remoting/capturer/video_frame_capturer_mac.mm:56: SkRect ScaleSkRect(const
SkRect& rect, float scale) {
I believe we already have a function like this somewhere in ui/gfx, but maybe
remoting doesn't depend on that

[Wez, 2013-01-24 00:57:30]

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
File remoting/capturer/mac/desktop_configuration.h (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.h:28: // Bounds of this display in
logical (72dpi) coordinates.
On 2013/01/24 00:26:58, Nico wrote:
> We call "logical coordinates" "dips" or "device-independent pixels" in
chromium
> code (following android's lead), and most things are in dips by default and
only
> stuff in pixels is called out as "pixel_foo".

Yes, I was going to describe these as DIPs but wanted to avoid confusion w/ web
content coordinate schemes.  It doesn't really matter; happy to change it to
DIPs here.

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.h:38: float logical_to_pixel_scale;
On 2013/01/24 00:26:58, Nico wrote:
> For what it's worth, we call this "device_scale" in most other places in
> chromium and webkit code.

Yes; I'm not keen on that name because we then end up with things like
Graphics2D::SetScale(1.0f / device_scale), which is hard to read - I think it's
best to be explicit as to the units between which the scale is expressed.

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.h:58: SkIPoint dpi;
On 2013/01/24 00:26:58, Nico wrote:
> I'm not sure if dpi is a useful metric on OS X, isn't it always 72?

We use DPI on-the-wire in Chromoting, but you're right, there's no need to deal
with DPI at this level.

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
File remoting/capturer/mac/desktop_configuration.mm (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.mm:58: [screen
respondsToSelector:@selector(convertRectToBacking)]) {
For some reason I'm getting this returning false...

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/v...
File remoting/capturer/video_frame_capturer_mac.mm (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/v...
remoting/capturer/video_frame_capturer_mac.mm:56: SkRect ScaleSkRect(const
SkRect& rect, float scale) {
On 2013/01/24 00:26:58, Nico wrote:
> I believe we already have a function like this somewhere in ui/gfx, but maybe
> remoting doesn't depend on that

I couldn't see anything in there that did a simple scale of an SkRect - the only
likely candidates looked to require converting to gfx::Rect and then applying a
matrix transform...

[Nico, 2013-01-24 00:58:30]

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
File remoting/capturer/mac/desktop_configuration.mm (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.mm:58: [screen
respondsToSelector:@selector(convertRectToBacking)]) {
On 2013/01/24 00:57:30, Wez wrote:
> For some reason I'm getting this returning false...

You're missing the trailing ':', it's "convertRectToBacking:"

[Wez, 2013-01-24 00:59:38]

On 2013/01/24 00:58:30, Nico wrote:
>
https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
> File remoting/capturer/mac/desktop_configuration.mm (right):
> 
>
https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
> remoting/capturer/mac/desktop_configuration.mm:58: [screen
> respondsToSelector:@selector(convertRectToBacking)]) {
> On 2013/01/24 00:57:30, Wez wrote:
> > For some reason I'm getting this returning false...
> 
> You're missing the trailing ':', it's "convertRectToBacking:"

*facepalm*

[Wez, 2013-01-25 00:32:49]

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
File remoting/capturer/mac/desktop_configuration.h (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.h:28: // Bounds of this display in
logical (72dpi) coordinates.
On 2013/01/24 00:57:30, Wez wrote:
> On 2013/01/24 00:26:58, Nico wrote:
> > We call "logical coordinates" "dips" or "device-independent pixels" in
> chromium
> > code (following android's lead), and most things are in dips by default and
> only
> > stuff in pixels is called out as "pixel_foo".
> 
> Yes, I was going to describe these as DIPs but wanted to avoid confusion w/
web
> content coordinate schemes.  It doesn't really matter; happy to change it to
> DIPs here.

Done.

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.h:58: SkIPoint dpi;
On 2013/01/24 00:57:30, Wez wrote:
> On 2013/01/24 00:26:58, Nico wrote:
> > I'm not sure if dpi is a useful metric on OS X, isn't it always 72?
> 
> We use DPI on-the-wire in Chromoting, but you're right, there's no need to
deal
> with DPI at this level.

Done.  Thanks for the suggestion!

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
File remoting/capturer/mac/desktop_configuration.mm (right):

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.mm:13: MAC_OS_X_VERSION_MAX_ALLOWED
< MAC_OS_X_VERSION_10_7
On 2013/01/24 00:20:18, Jamie wrote:
> Nit: indent the continuation line.

Done.

https://chromiumcodereview.appspot.com/12040058/diff/2002/remoting/capturer/m...
remoting/capturer/mac/desktop_configuration.mm:24: const int kDefaultDPI = 72;
On 2013/01/24 00:26:58, Nico wrote:
> nit: no indent in namespaces

Done.

[commit-bot: I haz the power, 2013-01-25 00:34:08]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/wez@chromium.org/12040058/4004

[commit-bot: I haz the power, 2013-01-25 05:08:10]

Change committed as 178761