> We can then describe the restriction on attributes as follows. It is an
> error if the TREX pattern after transformation into normal form has an
> "attribute" element with a descendant "ref" or "attribute" element.
I don't see any reason to mandate this constraint check. I'd rather want
to see this check done by "lint" like program.
Yes, attribute pattern containing attributes is a "broken" pattern, but
there are much more broken patterns. And we can't and shouldn't try to
detect all of them.
<define name="xxx" combine="choice">
...
</define>
For example, under the current specification, the above example is
considered OK even if no pattern is defined as "xxx". I guess this
feature is intentional to make XHTML modularization happy, but it is
probably not a good feature for humans, who tend to make typos.
--
Kohsuke KAWAGUCHI +1 650 786 0721
Sun Microsystems
