Bug 88263 – Add support for enabling text subpixel positioning in Skia

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKit Misc.

[Daniel Erat]

The Skia backend supports subpixel-positioned text (via SkPaint::setSubpixelText()), but we currently lack a way to enable it.  I have a change that adds a useSubpixelPositioning field to FontRenderStyle and WebFontRenderStyle and makes FontPlatformDataHarfBuzz apply it to the SkPaint.  It also renames the existing useSubpixel fields to useSubpixelRendering.

[Daniel Erat]

Created attachment 145665 [details]
Patch

[WebKit Review Bot]

Please wait for approval from [e-mail], [e-mail], [e-mail], [e-mail] or [e-mail] before submitting, as this patch contains changes to the Chromium public API. See also https://trac.webkit.org/wiki/ChromiumWebKitAPI.

[Behdad Esfahbod]

+    // FontConfig doesn't provide parameters to configure whether subpixel
+    // positioning should be used or not.
+    out->useSubpixelPositioning = 2;

A better default may be to enable this if hinting is off?

[WebKit Review Bot]

Comment on attachment 145665 [details]
Patch

Attachment 145665 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/12901496

New failing tests:
fonts/fantasy.html
fast/css/font-face-synthetic-bold-italic.html
fast/text/fake-italic.html
fonts/cursive.html
fast/css/font-face-multiple-faces.html

[WebKit Review Bot]

Created attachment 145778 [details]
Archive of layout-test-results from ec2-cr-linux-02

The attached test failures were seen while running run-webkit-tests on the chromium-ews.
Bot: ec2-cr-linux-02  Port: <class 'webkitpy.common.config.ports.ChromiumXVFBPort'>  Platform: Linux-2.6.35-28-virtual-x86_64-with-Ubuntu-10.10-maverick

[Daniel Erat]

#3: Not sure.  That seems like a fairly substantial change in behavior; I'd prefer for this change to be a no-op for now except in the case where we explicitly request subpixel positioning.

[Behdad Esfahbod]

Fair enough.  I'll go request an element for this in Fontconfig...

[Daniel Erat]

Created attachment 145823 [details]
Patch

[Daniel Erat]

The patch from #8 makes SimpleFontData::platformWidthForGlyph() again round widths, but now only when subpixel positioning is disabled.  It looks like SkPaint::measureText() is returning fractional widths for fake-italics (and maybe also fake-bold?).  Behdad suggests that Skia should be doing the rounding itself when subpixel positioning is disabled; I think that ensuring that this is the case within WebKit is safe.

[Tony Chang]

Comment on attachment 145823 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=145823&action=review

Code change looks fine.  You'll have to wait for an API reviewer to take a look at this change.

> Source/WebCore/ChangeLog:13
> +        Pass setting to SkPaint in FontPlatformDataHarfBuzz and remove
> +        round() call when computing glyph widths in SimpleFontDataSkia.
> +

Please add a sentence about why there are no new tests.  E.g., No new tests because this shouldn't change page rendering.  It might be possible to write a gtest style test (see the Source/WebKit/chromium/tests directory) for what is exposed via WebFontRenderStyle.h, but I'm not sure if you can verify that the value was properly set.

> Source/WebCore/platform/graphics/chromium/FontRenderStyle.h:49
> -          useSubpixel(0) { }
> +          useSubpixelRendering(0),
> +          useSubpixelPositioning(0) { }

Nit: WebKit style normally puts the , in front (lines up below the :).

[Daniel Erat]

Created attachment 145881 [details]
Patch

[Tony Chang]

Comment on attachment 145881 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=145881&action=review

> LayoutTests/platform/chromium-linux/fast/text/chromium-linux-text-subpixel-positioning.html:3
> +    <p>This test requires Chromium Linux <tt>test_shell</tt> in

Nit: I would change this text to "This test requires DumpRenderTree, as that ..."

[Daniel Erat]

Created attachment 145890 [details]
Patch

[James Robinson]

Comment on attachment 145890 [details]
Patch

Public WebKit API lgtm

[WebKit Review Bot]

Comment on attachment 145890 [details]
Patch

Attachment 145890 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/12897824

New failing tests:
platform/chromium-linux/fast/text/chromium-linux-text-subpixel-positioning.html

[WebKit Review Bot]

Created attachment 146030 [details]
Archive of layout-test-results from ec2-cr-linux-02

The attached test failures were seen while running run-webkit-tests on the chromium-ews.
Bot: ec2-cr-linux-02  Port: <class 'webkitpy.common.config.ports.ChromiumXVFBPort'>  Platform: Linux-2.6.35-28-virtual-x86_64-with-Ubuntu-10.10-maverick

[Daniel Erat]

Created attachment 146119 [details]
Patch

[Tony Chang]

Comment on attachment 146119 [details]
Patch

Adding a new entry to fonts.conf seems fine.

[WebKit Review Bot]

Comment on attachment 146119 [details]
Patch

Clearing flags on attachment: 146119

Committed r119732: <http://trac.webkit.org/changeset/119732>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Daniel Erat]

Reopening to attach new patch.

[Daniel Erat]

Created attachment 147334 [details]
Patch

[Daniel Erat]

The new patch removes a field that was deprecated by this change.

[Adam Barth]

(In reply to comment #21)
> Reopening to attach new patch.

Normally we use a new bug for each patch.

[WebKit Review Bot]

Comment on attachment 147334 [details]
Patch

Clearing flags on attachment: 147334

Committed r120363: <http://trac.webkit.org/changeset/120363>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.