Set up relations described on https://wiki.mozilla.org/Accessibility/MathML.

Added all relations listed on the wiki page. The necessity of these relations is still in discussion (forward direction, e.g. fraction -> numerator, due to the rigidity of the MathML spec, and the backwards direction, e.g. numerator -> fraction, due to various other factors).

Switched to utils class for common MathML implementations.

Comment on attachment 8434644 [details] [diff] [review] MathML relations Review of attachment 8434644 [details] [diff] [review]: ----------------------------------------------------------------- rs=me ::: accessible/src/generic/MathMLUtils.cpp @@ +534,5 @@ > + case RelationType::MATHML_ROW_FOR: { > + if (accRole != roles::MATHML_ROW) > + break; > + > + aRelation->AppendIter(new ChildIterator(aAccessible)); this is rather crazy since its full analogue of GetChildren() but that may be ok for now. We'd need to run all of these through some implementation to see what makes sense and what should go away

Unbitrotting...

Unbitrot

Unbitrotting the patch again. However, the way we map MathML elements to roles has changed in Alexander's patch for bug 1001635, so this might need some refactoring...

yes, I'll handle it

This is the continuation of attachment 8623966 [details] [diff] [review] to move relations into MarkupMap. IIUC, if we have <mfrac> [NUMERATOR] [DENOMINATOR] </mfrac> then we can implement the relation RELATION_MATHML_NUMERATOR by associating the relation to the nsGkAtoms::mfrac_ element. However, to implement the converse relation RELATION_MATHML_DENOMINATOR_OF, we would need to associate it to *any* MathML elements that can potentially be a numerator. So I don't think this map stuff is really a good idea to implement Jonathan's relations.