The Unicode spec define some mirroring per character:
The attached patch (which applies on top of bug 130322) does the necessary changes to mirror these characters for stretchy Operators drawn with the MATH table. This is just using the appropriate "mirror" boolean when calling glyphDataForCharacter on the base glyph.
It's not really possible to do the same for the old code since for example the parenthesis pieces are not defined as mirrorable in the Unicode spec. However, there are not many fences in the stretchyCharacters table, so it is certainly possible to add some ad hoc code to mirror the base character. But perhaps it's not worth it if we want to move to MATH fonts.
For other operators like sums or integrals, we will need "Glyph-level mirroring" as well as fonts such as XITS that contain the mirrored glyphs:
However, I'm not sure the 'rtlm' feature is implemented in WebKit so I'll open a separate bug for that.
Created attachment 227946 [details]
Created attachment 228830 [details]
Created attachment 228831 [details]
Patch+130322 for testing
Created attachment 232528 [details]
Committed r169617: <http://trac.webkit.org/changeset/169617>
Comment on attachment 232528 [details]
View in context: https://bugs.webkit.org/attachment.cgi?id=232528&action=review
> + index += index % 2 ? -1 : 1;
There is likely a bug here. The new value assigned to index is never used because of the break below. Frederic, could you please check? I am not sure what was the intent because the chunk added here has no effect.
(In reply to comment #6)
> Comment on attachment 232528 [details]
> View in context:
> > Source/WebCore/rendering/mathml/RenderMathMLOperator.cpp:1587
> > + index += index % 2 ? -1 : 1;
> There is likely a bug here. The new value assigned to index is never used
> because of the break below. Frederic, could you please check? I am not sure
> what was the intent because the chunk added here has no effect.
I think the index change should have been done before the
stretchyCharacter = &stretchyCharacters[index];