Gate subpixel scaling by RuntimeEnabledFeatures::subpixelFontScalingEnabled() on Mac

Gate subpixel scaling by RuntimeEnabledFeatures::subpixelFontScalingEnabled() on Mac

Synchronise releasing / default activating this feature with the other platforms.

BUG=378195
R=eae,dglazkov,eseidel

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=175203

[Dominik Röttsches, 2014-05-28 14:33:51]

https://codereview.chromium.org/304723002/diff/1/Source/platform/fonts/mac/Si...
File Source/platform/fonts/mac/SimpleFontDataMac.mm (right):

https://codereview.chromium.org/304723002/diff/1/Source/platform/fonts/mac/Si...
Source/platform/fonts/mac/SimpleFontDataMac.mm:329:
scaledFontData.setFont(useHinting() ? [scaledFontData.font() screenFont] :
[scaledFontData.font() printerFont]);
This part should have been in the previous patch for removing
setUsePrinterFont() removal.

[eae, 2014-05-28 16:15:05]

This is great! 

Could you also unskip the fast/text/sub-pixel/text-scaling* tests for Mac with
this patch or are further changes needed? See line 365-369 in TestExpectations.

[eae, 2014-05-28 19:06:11]

LGTM (but please unskip the text scaling tests if you can).

[eae, 2014-05-31 19:02:55]

The CQ bit was checked by eae@chromium.org

[commit-bot: I haz the power, 2014-05-31 19:03:29]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/dominik.rottsches@intel.com/304723002/1

[commit-bot: I haz the power, 2014-05-31 20:28:25]

Change committed as 175203

[eae, 2014-05-31 23:27:52]

On 2014/05/31 20:28:25, I haz the power (commit-bot) wrote:
> Change committed as 175203

Enabling tests in https://codereview.chromium.org/306243002

[Dominik Röttsches, 2014-06-01 17:49:59]

> Enabling tests in https://codereview.chromium.org/306243002

Thanks for landing! Saw the comment regarding the tests.

[eae, 2014-06-10 23:13:16]

It appears this change only effects system fonts and not web fonts. Perhaps we
need to do the same dance in FontCustomPlatformDataMac.cpp?

[Dominik Röttsches, 2014-06-11 07:40:15]

On 2014/06/10 23:13:16, eae wrote:
> It appears this change only effects system fonts and not web fonts. Perhaps we
> need to do the same dance in FontCustomPlatformDataMac.cpp?

Can you clarify what you mean by "affects"? Do you mean hinting is still enabled
for custom fonts, initialized through FontCustomPlatformDataMac?

I believe the issues that in FontCustomPlatformDataMac fonts are intialized
through CoreGraphics API, the way it should be, while our other system fonts are
instantiated through AppKit - which we should get rid of (crbug.com/364983). As
far as I understand, CGFonts decide whether to use hinting or not based on the
surface, whereas in NSFonts it's configured in the font.

[eae, 2014-06-13 18:05:09]

On 2014/06/11 07:40:15, Dominik Röttsches wrote:
> On 2014/06/10 23:13:16, eae wrote:
> > It appears this change only effects system fonts and not web fonts. Perhaps
we
> > need to do the same dance in FontCustomPlatformDataMac.cpp?
> 
> Can you clarify what you mean by "affects"? Do you mean hinting is still
enabled
> for custom fonts, initialized through FontCustomPlatformDataMac?

Compare http://blink.gs/tests/text-scaling/webfont.html and
http://blink.gs/tests/text-scaling/ltr.html on mac. On all other platforms
webfonts scale smoothly just like system fonts, on mac not so much.

[Dominik Röttsches, 2014-06-17 13:34:27]

On 2014/06/13 18:05:09, eae wrote:
> On 2014/06/11 07:40:15, Dominik Röttsches wrote:
> > On 2014/06/10 23:13:16, eae wrote:
> > > It appears this change only effects system fonts and not web fonts.
Perhaps
> we
> > > need to do the same dance in FontCustomPlatformDataMac.cpp?
> > 
> > Can you clarify what you mean by "affects"? Do you mean hinting is still
> enabled
> > for custom fonts, initialized through FontCustomPlatformDataMac?
> 
> Compare http://blink.gs/tests/text-scaling/webfont.html and
> http://blink.gs/tests/text-scaling/ltr.html on mac. On all other platforms
> webfonts scale smoothly just like system fonts, on mac not so much.

Thanks for the test links. Seems to me that the metrics and the (subpixel) glyph
positioning in theses cases are actually okay, but the glyph sizes are somehow
quantized. I am currently suspecting the Skia glyph bitmap creation to do
something different for custom fonts. Need to investigate more.

[eae, 2014-06-17 18:14:49]

On 2014/06/17 13:34:27, Dominik Röttsches wrote:
> On 2014/06/13 18:05:09, eae wrote:
> > On 2014/06/11 07:40:15, Dominik Röttsches wrote:
> > > On 2014/06/10 23:13:16, eae wrote:
> > > > It appears this change only effects system fonts and not web fonts.
> Perhaps
> > we
> > > > need to do the same dance in FontCustomPlatformDataMac.cpp?
> > > 
> > > Can you clarify what you mean by "affects"? Do you mean hinting is still
> > enabled
> > > for custom fonts, initialized through FontCustomPlatformDataMac?
> > 
> > Compare http://blink.gs/tests/text-scaling/webfont.html and
> > http://blink.gs/tests/text-scaling/ltr.html on mac. On all other platforms
> > webfonts scale smoothly just like system fonts, on mac not so much.
> 
> Thanks for the test links. Seems to me that the metrics and the (subpixel)
glyph
> positioning in theses cases are actually okay, but the glyph sizes are somehow
> quantized. I am currently suspecting the Skia glyph bitmap creation to do
> something different for custom fonts. Need to investigate more.

Yeah it seems like the sizing is off but the individual glyphs render fine,
interestingly the results aren't consistent, reloading the page sometimes
results in a very different rendering.

[Dominik Röttsches, 2014-06-19 09:10:43]

I did a small Skia test based on SkiaExamples and it seems, at least in the
example canvas setup, Skia renders subpixel fonts with LCD antialiasing just
fine.

[eae, 2014-06-23 15:56:08]

On 2014/06/19 09:10:43, Dominik Röttsches wrote:
> I did a small Skia test based on SkiaExamples and it seems, at least in the
> example canvas setup, Skia renders subpixel fonts with LCD antialiasing just
> fine.

Odd, I wonder why we get a different behavior for webfonts then.

[Dominik Röttsches, 2014-07-23 07:53:44]

On 2014/06/23 15:56:08, eae wrote:
> Odd, I wonder why we get a different behavior for webfonts then.

This might explain it:
http://code.google.com/p/chromium/issues/detail?id=395090#c12

[eae, 2014-07-23 14:52:30]

On Wed, Jul 23, 2014 at 12:53 AM,  <dominik.rottsches@intel.com> wrote:
> On 2014/06/23 15:56:08, eae wrote:
>>
>> Odd, I wonder why we get a different behavior for webfonts then.
>
>
> This might explain it:
> http://code.google.com/p/chromium/issues/detail?id=395090#c12

That is only for linux though (and it works for webfonts on linux).
Perhaps we do something similar on mac?

To unsubscribe from this group and stop receiving emails from it, send an email
to blink-reviews+unsubscribe@chromium.org.