Re: [David Carlisle] Re: question about xlink 1.1

> From: David Carlisle <davidc@nag.co.uk>
[...]
>
> http://lists.w3.org/Archives/Public/public-xml-core-wg/2009Jun/0030.html
>
>> However, I am left with one residual query in this area:  why isn't
>> href #REQUIRED in simpleAttrs in the DTD (and schemas)?  Why does the
>> spec. say "it MAY be used on simple-type elements"?  What does a
>> simple link w/o an href mean?
>
> a dtd fragment using #IMPLIED  is more useful than one using #REQUIRED
> for many uses of xlink (and for similar reasons the dtd shouldn't
> default a type="simple" attribute, now that xlink 1.1 doesn't require
> one.
>
> Consider for example MathML2 where every element can be used as a link,
> by using xlink:href attribute. It's useful to be able to include a dtd
> fragment that sets up #IMPLIED xlink attributes so that this usage is
> DTD valid but you can't _require_ href on every element.

Right.

> Defaulting type="simple" is slightly different. xlink 1.0 mandated use
> of type attribute so mathml originally defaulted type="simple" in the
> dtd which looked fine from an authoring point of view, but we got very
> negative feedback from mozilla (in particular) because if they used a
> dtd that defaulted this attribute it was really there in the DOM on
> every element and in a markup heavy language like mathml that is a heavy
> cost, especially if the majority of documents are not using xlink at
> all.  So at some point we changed the dtd not to default type="simple"
> which meant it had to be explict on every link. So xlink 1.1 not
> requiring this would be a good thing, but not using a dtd that defaults
> the attribute again.
>
> Feel free to forward or discard as you see fit:-)

Also right, I think.


                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | The most violent element in society is
http://nwalsh.com/            | ignorance.--Emma Goldman