Use text size on Mac.

Use text size on Mac.

The current code assumes that text on Mac is freely scalable,
or at least that the text size is just a part of the transform.
However, it appears that application of the 'trak' table,
as well as other optical adjustments, may rely on the text size
directly.

This change passes the text size requested directly to CoreText.

BUG=chromium:427528
Committed: https://skia.googlesource.com/skia/+/be2284de556e7ec29489693093d42dc86b762645

[bungeman-skia, 2014-11-24 22:45:47]

bungeman@google.com changed reviewers:
+ mtklein@google.com, reed@google.com

[bungeman-skia, 2014-11-24 22:45:48]

Currently testing to see what this might do to the Blink layout tests at
https://codereview.chromium.org/756743004 . So don't go CQ'ing this until we
know.

[bungeman-skia, 2014-11-25 02:14:57]

On 2014/11/24 22:45:48, bungeman1 wrote:
> Currently testing to see what this might do to the Blink layout tests at
> https://codereview.chromium.org/756743004 . So don't go CQ'ing this until we
> know.

At Patch Set 3 this passes the our tests and Blink's layout tests. I'd create a
test, but I think we need an OS X 10.10 bot for it.

[reed1, 2014-11-25 14:31:35]

lgtm w/ request for slight refactorying/comments now or in a follow-up CL.

https://codereview.chromium.org/752183002/diff/40001/src/ports/SkFontHost_mac...
File src/ports/SkFontHost_mac.cpp (right):

https://codereview.chromium.org/752183002/diff/40001/src/ports/SkFontHost_mac...
src/ports/SkFontHost_mac.cpp:720: SkMatrix skTransform;
Maybe capture this in a helper function, like

static CGAffineTransform compute_transform(SkScalerContext::Rec* rec) ...

That might help isolate it, so I am reminded to offer this sort of computation
directly on Rec (without needing to mutate it as you have to here).

https://codereview.chromium.org/752183002/diff/40001/src/ports/SkFontHost_mac...
src/ports/SkFontHost_mac.cpp:760: fRec.getSingleMatrix(&fFUnitMatrix);
// Here we don't swap-in a 1.0 for textsize, but that's fine because ...

[bungeman-skia, 2014-11-25 15:36:11]

How about like this?

https://codereview.chromium.org/752183002/diff/40001/src/ports/SkFontHost_mac...
File src/ports/SkFontHost_mac.cpp (right):

https://codereview.chromium.org/752183002/diff/40001/src/ports/SkFontHost_mac...
src/ports/SkFontHost_mac.cpp:720: SkMatrix skTransform;
On 2014/11/25 14:31:35, reed1 wrote:
> Maybe capture this in a helper function, like
> 
> static CGAffineTransform compute_transform(SkScalerContext::Rec* rec) ...
> 
> That might help isolate it, so I am reminded to offer this sort of computation
> directly on Rec (without needing to mutate it as you have to here).

Added to Rec.

https://codereview.chromium.org/752183002/diff/40001/src/ports/SkFontHost_mac...
src/ports/SkFontHost_mac.cpp:760: fRec.getSingleMatrix(&fFUnitMatrix);
On 2014/11/25 14:31:35, reed1 wrote:
> // Here we don't swap-in a 1.0 for textsize, but that's fine because ...

Done.

[reed1, 2014-11-25 15:41:30]

purty
lgtm

[bungeman-skia, 2014-11-25 15:56:27]

The CQ bit was checked by bungeman@google.com

[commit-bot: I haz the power, 2014-11-25 15:57:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/752183002/60001

[commit-bot: I haz the power, 2014-11-25 16:08:12]

Committed patchset #4 (id:60001) as
https://skia.googlesource.com/skia/+/be2284de556e7ec29489693093d42dc86b762645