Use more fields from OutlineFontMEtrics: linegap, bbox.

Use more fields from OutlineFontMEtrics: linegap, bbox.
Up our canonical size to 2048 to reduce precision loss.

R=bungeman@google.com

Committed: https://code.google.com/p/skia/source/detail?r=9057

[reed1, 2013-05-07 19:22:35]

causes minor gm failures/rebaselines. Don't know its impact on DRT yet

[bungeman-skia, 2013-05-07 19:56:08]

https://codereview.chromium.org/14759021/diff/1/src/ports/SkFontHost_win.cpp
File src/ports/SkFontHost_win.cpp (right):

https://codereview.chromium.org/14759021/diff/1/src/ports/SkFontHost_win.cpp#...
src/ports/SkFontHost_win.cpp:106: static const int gCanonicalTextSize = 2048;
Ick! Ah well, it used to be the magic number 64, this number isn't much worse.

https://codereview.chromium.org/14759021/diff/1/src/ports/SkFontHost_win.cpp#...
src/ports/SkFontHost_win.cpp:840: void
SkScalerContext_Windows::generateFontMetrics(SkPaint::FontMetrics* mx,
SkPaint::FontMetrics* my) {
Just wanted to comment here so we don't forget. Having requests here for the two
different types (vertical and horizontal metrics) is a bit silly, since the
scaler context is already either vertical or horizontal.

https://codereview.chromium.org/14759021/diff/1/src/ports/SkFontHost_win.cpp#...
src/ports/SkFontHost_win.cpp:841: // Note: This code was borrowed from
generateLineHeight, which has a note
Now that there is nothing left of the original code, perhaps we can remove this
comment.

https://codereview.chromium.org/14759021/diff/1/src/ports/SkFontHost_win.cpp#...
src/ports/SkFontHost_win.cpp:895: my->fAscent = -fScale * otm.otmAscent;
This ascent, descent, and leading look like they match what
Source/core/platform/graphics/chromium/SimpleFontDataChromiumWin.cpp does. They
also want a valid x-height, average and max character width (though I have no
idea what use they might be).

[reed1, 2013-05-07 21:04:48]

https://codereview.chromium.org/14759021/diff/1/src/ports/SkFontHost_win.cpp
File src/ports/SkFontHost_win.cpp (right):

https://codereview.chromium.org/14759021/diff/1/src/ports/SkFontHost_win.cpp#...
src/ports/SkFontHost_win.cpp:840: void
SkScalerContext_Windows::generateFontMetrics(SkPaint::FontMetrics* mx,
SkPaint::FontMetrics* my) {
On 2013/05/07 19:56:08, bungeman1 wrote:
> Just wanted to comment here so we don't forget. Having requests here for the
two
> different types (vertical and horizontal metrics) is a bit silly, since the
> scaler context is already either vertical or horizontal.

That CL is on the way

https://codereview.chromium.org/14759021/diff/1/src/ports/SkFontHost_win.cpp#...
src/ports/SkFontHost_win.cpp:841: // Note: This code was borrowed from
generateLineHeight, which has a note
On 2013/05/07 19:56:08, bungeman1 wrote:
> Now that there is nothing left of the original code, perhaps we can remove
this
> comment.

Done.

https://codereview.chromium.org/14759021/diff/1/src/ports/SkFontHost_win.cpp#...
src/ports/SkFontHost_win.cpp:895: my->fAscent = -fScale * otm.otmAscent;
On 2013/05/07 19:56:08, bungeman1 wrote:
> This ascent, descent, and leading look like they match what
> Source/core/platform/graphics/chromium/SimpleFontDataChromiumWin.cpp does.
They
> also want a valid x-height, average and max character width (though I have no
> idea what use they might be).

x-height and avecharwidth -- done

SkFontMetrics doesn't have a field for max-char-width

[bungeman-skia, 2013-05-08 02:40:40]

Other than some moaning about bitmap fonts, lgtm.

https://codereview.chromium.org/14759021/diff/6001/src/ports/SkFontHost_win.cpp
File src/ports/SkFontHost_win.cpp (right):

https://codereview.chromium.org/14759021/diff/6001/src/ports/SkFontHost_win.c...
src/ports/SkFontHost_win.cpp:868: my->fLeading =
SkIntToScalar(fTM.tmExternalLeading);
tmAveCharWidth is also available here. There is no xheight, however. Blink
currently uses tmAscent * 0.56f as a guess.

Apparently the avg and max character widths are used to size text boxes. The
xheight is semi-optional and is mostly used as the basis of the css ex unit, but
is also used to attempt vertical centering.

[reed1, 2013-05-08 14:11:31]

Committed patchset #2 manually as r9057 (presubmit successful).