Print

Print


> Originally, for example, abbr/expan were paired thusly:
> <abbr expan="um">&umacr;</abbr> (which is fine) and <expan
> abbr="&umacr;">um</expan> (which is not, since attribute values are not
> parsed).

Not so, I fear.  There is no reason why references (whether they are entity
references or numeric ones) shouldn't be *used* in attribute values. The
only restriction is that they must not *expand to* anything (such as markup)
which is illegal in an attribute value. So provided there was an external or
internal subset which assigned a legal expansion to &umacr; then <expan>
abbr="&umacr;">um</expan> would itself be perfectly legal.

What may be the cause of the confusion here is that references can never be
used in the names of elements or attributes in XML. But that doesn't arise
with Janus tags. The other potential source of confusion is the problem
which occurred with such examples if the CER  expanded either to something
that is not a legal Unicode codepoint, or to something (such as a <c> or
[TADAA!!] a brand new P5 <g> which entails markup. Now that is something
which Janus-ing can't cope with but <choice> can, since element content can
contain any markup the content model allows at that point.

Michael Beddow