> But, of course, piggy-backing PIs onto elements using attributes
> is probably always a preferable option, if your PI structure can
> matches the start-tag locations.
Do you mean that instead of using:
<?XSchema SYSTEM="foo.xsc" ?>
<foo>
...
</foo>
we should use something like:
<foo XSchemaSystem="foo.xsc">
...
</foo>
A good idea, but I don't think it will work for us. We accept an arbitrary
number of XSchema PI's, so there is no way to name the attributes.
>
> Even if you use attribute-like syntax, PI values are still just
> text strings, from the XML viewpoint. Having a target (notation)
> to indicate the syntax of the PI is a great headstart in labelling.
I don't understand what you mean. A PI has a target by definition, but I don't
know how this indicates syntax.
-- Ron Bourret