W3C home > Mailing lists > Public > www-math@w3.org > August 2007

Re: italic dotless letters

From: William F Hammond <hammond@csc.albany.edu>
Date: Tue, 21 Aug 2007 14:59:38 -0400
To: www-math@w3.org
Message-ID: <i764384sth.fsf@hilbert.math.albany.edu>

Richard Kaye <R.W.Kaye@bham.ac.uk> writes in part:

> Suppose I write a MathML2.0 document, using these entities.  In 
> a few years time new unicode characters may be introduced,
> or the wrinkles in MathML DTDs may be ironed out, and these
> entities are changed.  Then my document suddenly doesn't display
> correctly with the new version of MathML, and the required
> changes are highly non-obvious.  This wouldn't be a good state
> of affairs ...

(I'm ignoring what was said later.)

IHMO the XHTML+MathML spun out by an authoring interface should
not use these names but rather use the corresponding hexadecimal
cdata character entities, e.g., "&#x220F;".  Moreover it is
unsatisfactory to have any form of cdata character entity in
a format that is transformed to XHTML+MathML because conforming
translators will convert them to the actual live characters and
that will break some current rendering agents (not always their
fault -- so be patient).

It's certainly important to have names in authoring interfaces.  A
limited number of such names can be made available as empty elements
(but then they must be fit structurally into the document type
definition, which is a burden but not a huge burden).  Or we could
petition higher powers to allow SDATA entities in XML, version > 1.
The difference is that, while they need definition when there is a
document type definition, they do not need to be fit into document
structure (other than where normal characters are allowed to go).

There is even a sane way to understand XML documents with SDATA
entities as "standalone".  In that regard (undefined) SDATA entities
in a standalone XML document would be parallel to empty elements.  In
both cases a rendering agent needs to recognize the names.  In fact,
as things are now, we apparently have at least one rendering agent
that in math zones ports many of these characters back to the names
for internal handling -- nearly a de facto re-introduction of SDATA.


                                    -- Bill
Received on Tuesday, 21 August 2007 18:59:41 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:27:39 UTC