Work around changes in OSX 10.10.

Work around changes in OSX 10.10.

In 10.10.1 CTFontGetAdvancesForGlyphs applies the font
transform to the width of the advance but always sets the
height of the advance to 0. Work around this by measuring the
advance with an unrotated font and apply the transform
manually.

In 10.7 through 10.9 CTFontDrawGlyphs appears to take
'positions' in text space and acts like
CGContextShowGlyphsAtPosition. In 10.10.1, these 'positions' now appear to be in glyph space, and
CTFontDrawGlyphs maps the glyph positions internally.

BUG=chromium:442574
Committed: https://skia.googlesource.com/skia/+/3b4b66c04a2567c72754a55e53db41394249b223

[bungeman-skia, 2015-01-08 00:11:24]

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

[bungeman-skia, 2015-01-08 00:11:24]

Will add more to description later, my editor eated it.

[mtklein, 2015-01-08 14:45:54]

I think I follow.

https://codereview.chromium.org/841843002/diff/1/src/ports/SkFontHost_mac.cpp
File src/ports/SkFontHost_mac.cpp (right):

https://codereview.chromium.org/841843002/diff/1/src/ports/SkFontHost_mac.cpp...
src/ports/SkFontHost_mac.cpp:49: #define SK_SHOW_TEXT_BLIT_COVERAGE 0
Add a little note about what's going on here?  Is this something like //
Brighten all components to a minimum so coverage is more visible?

https://codereview.chromium.org/841843002/diff/1/src/ports/SkFontHost_mac.cpp...
src/ports/SkFontHost_mac.cpp:905: // Prior to 10.10, CTFontDrawGlyphs and (the
no longer used) CGContextShowGlyphsAtPositions
might drop off the bit about no-longer-used code?

https://codereview.chromium.org/841843002/diff/1/src/ports/SkFontHost_mac.cpp...
src/ports/SkFontHost_mac.cpp:1019: // Vertical metrics are rotated 90 degrees.
... 90 degrees clockwise.  ?

[reed1, 2015-01-08 15:28:02]

time to switch to freetype...

lgtm

[bungeman-skia, 2015-01-08 16:21:58]

https://codereview.chromium.org/841843002/diff/1/src/ports/SkFontHost_mac.cpp
File src/ports/SkFontHost_mac.cpp (right):

https://codereview.chromium.org/841843002/diff/1/src/ports/SkFontHost_mac.cpp...
src/ports/SkFontHost_mac.cpp:49: #define SK_SHOW_TEXT_BLIT_COVERAGE 0
On 2015/01/08 14:45:54, mtklein wrote:
> Add a little note about what's going on here?  Is this something like //
> Brighten all components to a minimum so coverage is more visible?

Done. Hmmm... I keep adding this without explanation as it becomes useful.
FreeType and GDI both already have this (with the same name). I should add this
to DirectWrite and move this define out to SkUserConfig. On the other hand it's
nice to control it within the compilation unit as that means faster cycle times
to turning it on, and also I can turn it on for GDI but not for DirectWrite so I
can tell who is drawing with what. But yeah, this is just a debugging aid which
makes the bounding box visible. With most formats it forces the alpha to some
minimum value, with black and white I normally put dots in the corners.

https://codereview.chromium.org/841843002/diff/1/src/ports/SkFontHost_mac.cpp...
src/ports/SkFontHost_mac.cpp:905: // Prior to 10.10, CTFontDrawGlyphs and (the
no longer used) CGContextShowGlyphsAtPositions
On 2015/01/08 14:45:54, mtklein wrote:
> might drop off the bit about no-longer-used code?

Done. Thanks for reminding me about this comment. After some investigation, it
turns out the behavior of CGContextShowGlyphsAtPositions has actually not
changed, and seems to still have the old behavior.

https://codereview.chromium.org/841843002/diff/1/src/ports/SkFontHost_mac.cpp...
src/ports/SkFontHost_mac.cpp:1019: // Vertical metrics are rotated 90 degrees.
On 2015/01/08 14:45:54, mtklein wrote:
> ... 90 degrees clockwise.  ?

Done. As if they were horizontal. They're not really rotated, they just stick
the advance in the width instead of the height. I hadn't really thought about it
that way until I just tried to decide if it was rotated clockwise or not.

[bungeman-skia, 2015-01-08 16:22:49]

The CQ bit was checked by bungeman@google.com

[commit-bot: I haz the power, 2015-01-08 16:23:39]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/841843002/20001

[commit-bot: I haz the power, 2015-01-08 16:33:47]

Committed patchset #2 (id:20001) as
https://skia.googlesource.com/skia/+/3b4b66c04a2567c72754a55e53db41394249b223

[behdad_google, 2015-01-08 18:16:36]

behdad@google.com changed reviewers:
+ behdad@google.com

[behdad_google, 2015-01-08 18:16:36]

What are the 'lum' changes doing in this CL?