>So I don't think elements provide what PIs do: if elements were
satisfactory, people would use elements.
We had an interesting debate during XSD 1.1 development on whether PIs
(and comments) should be visible to assertions. On the one hand, people
felt there should be a way for the schema to control whether PIs were
allowed or not. But other people felt that the whole point of PIs is
that enable document authors and recipients to get around restrictions
imposed by unimaginative schema designers.
The final outcome was: By default, comments and processing instructions
are excluded from the partial ·post-schema-validation infoset· [visible
to an assertion], but ·at user option· processors may retain comments
and processing instructions instead of excluding them. (Saxon supplies
such an option.)
Michael Kay
Saxonica
