Bug 114999 – Disable Subpixel layout on mac

• Component: WebKit Misc.
• Status: RESOLVED FIXED

[Roger Fong]

<rdar://problem/13667607>

[Roger Fong]

Created attachment 199335 [details]
patch

[zalan]

+@@ -256,7 +256,7 @@
+ #endif
+ 
+ #if !defined(ENABLE_SUBPIXEL_LAYOUT)
+-#define ENABLE_SUBPIXEL_LAYOUT 1
++#define ENABLE_SUBPIXEL_LAYOUT 0
+ #endif
+ 
+ #endif /* PLATFORM(EFL) */

This is EFL. I don't think it is applicable.

[Roger Fong]

Oops, yeah, wrong place. Meant to do the one near the end of the file.

[zalan]

(In reply to comment #3)
> Oops, yeah, wrong place. Meant to do the one near the end of the file.
I dont think that needs to be changed either.

[Roger Fong]

Created attachment 199771 [details]
patch

Got rid of the unnecessary define.
This is just to disable sub pixel layout. 
The test rebaselines will come in a separate patch.

[Roger Fong]

Created attachment 199772 [details]
test rebaselines

test reaselines.
Note there’s there’s still a couple dozen failures that I plan on skipping and dealing with separately after this lands.

[zalan]

Comment on attachment 199771 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=199771&action=review

> WebCore/platform/graphics/LayoutRect.h:224
>  #endif

Any particular reason why you switched from 'explicit IntRect(const LayoutRect&)' to 'IntRect(int x, int y, int width, int height)'?

[Allan Sandfeld Jensen]

Since Qt and GTK enabled sub-pixel layout last week, it would be interesting if you could share some of the reasoning for Mac disabling it again.

[zalan]

(In reply to comment #8)
> Since Qt and GTK enabled sub-pixel layout last week, it would be interesting if you could share some of the reasoning for Mac disabling it again.

Various off-by-one pixel issues as we just discussed it at #qtwebkit.

[Roger Fong]

(In reply to comment #7)
> (From update of attachment 199771 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=199771&action=review
> 
> > WebCore/platform/graphics/LayoutRect.h:224
> >  #endif
> 
> Any particular reason why you switched from 'explicit IntRect(const LayoutRect&)' to 'IntRect(int x, int y, int width, int height)'?

Build fix. Not quite sure why the old one stopped working. I'll look into it.

[Roger Fong]

Created attachment 199855 [details]
patch

Change build fix to add an export

[Roger Fong]

Created attachment 199857 [details]
Patch

[Simon Fraser (smfr)]

Comment on attachment 199857 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=199857&action=review

> Source/WebCore/ChangeLog:9
> +        Reviewed by NOBODY (OOPS!).
> +
> +        * Configurations/FeatureDefines.xcconfig:
> +        * WebCore.exp.in:

This should say why you changed WebCore.exp.in

[Roger Fong]

committed: http://trac.webkit.org/changeset/149209

[zalan]

yay!

[Balazs Kelemen]

(In reply to comment #9)
> (In reply to comment #8)
> > Since Qt and GTK enabled sub-pixel layout last week, it would be interesting if you could share some of the reasoning for Mac disabling it again.
> 
> Various off-by-one pixel issues as we just discussed it at #qtwebkit.

This is a great source of platform bugs/misbehavior. I absolutely incompetent in the topic but it seems to me that diverging across ports in such a fundamental behavior is unhealthy. I think all platforms should make a consensus on this otherwise we are developing different engines in the same project which does not make sense.