Re: Directionality attribute

Thanks for your review.

Some personal responses, not confirmed with the working group.

> - In section 3.2.2, it is indicated that "MathML includes two 
> mathematics style attributes as well as a directionality attribute valid 
> on all presentation token elements, and on no other elements except 
> mstyle.", but apparently they are all valid on the <math/> element 

Thank you for pointing this out, I suspect the test was true when
written but in the end the current draft went with allowing all mstyle
attributes on math, rather than maintaining a separate list of
attributes that could be set on math, which was becoming increrasingly
ad hoc and hard to justify.  I would suggest that we add "and math"
after  "no other elements except mstyle"


> - Moreover, if I read the RelaxNG schema correctly, the dir attribute is 
> included in TokenAtt, itself included in mglyph.deprecatedattributes.

MathML2 defined mglyph as a token attribute but this was always rather
anomolous since it didn't take content, and appeared as children of
token elements. MathML3 reclassifies this as not being a token element.

3.2.1.2.4  says
> ... to make clear that mglyph is not a token element, ...

The intention was to allow but deprecate the attributes that were
allowed on all token elements. However the way it's done in the schema
by including TokenAtt into mglyph.deprecatedattributes does have the
affect of allowing dir as you say. Either we should say dir is allowed
but deprecated (which has a kind of logic, but may be strange to
deprecate a new attribute) or the schema should not include TokenAtt
but instead include each of the tokem attributes except for dir
individually. Something for the WG to decide.

> - Finally, it seems that allowing dir on the token element <mspace/> is 
> useless, because this element is empty.

the fact that it's empty doesn't necessarily mean that it doesn't have
an effect, in this case it could (perhaps, I think) affect linebreaking
choices (although I don't think any currently specified linebreaking
behaviour would be affected). Perhaps more realisticly if we do decide
to keep dir on the empty mglyph element then it could potentially affect
the rendering of glyphs from fonts that are subject to mirroring in a
bidirectional context.

 Currently dir is added automatically by the schema generator following
automatically from the fact that mspace is classified as a token
attribute. I think (personally) it is simpler to allow the attribute
even if there is no obvious use for it rather than say everywhere that
we discuss attributes on token elements "except mspace".



David

Received on Monday, 4 January 2010 12:23:26 UTC