the extent of the prefix binding

I. please specify the extent of the prefix binding as well as the scope

while the wd is clear that the scope of a prefix-to-uri binding within an
element is dynamic, the intended extent of the binding is unclear. the
discussion which has appeared on xml-dev, for example, leaves room for the
interpretation, that the binding has indefinite extent.

a binding which is afforded dynamic extent has a clearly understandable
semantics and is readily implemented, since the parser's state is sufficient
to implement the "inheritance". it would even be possible to implement it on
the basis of "ephemeral" attributes - which a serializer introduced into the
stream on the fly as needed and which a non validating parser would be free to
discard after application.

if the binding is intended to have an indefinite extent, which interpretation
some xml-dev discussion might be taken to imply, (see, for example, mr clark's
remark on passing state between the parser and the application for the purpose
of decoding attribute values:
http://www.lists.ic.ac.uk/hypermail/xml-dev/9808/0173.html) then an
implementation which retains a defined semantics in the presence of side
effects is more difficult (it would appear analogous to the "upward funarg" problem).

since the benefit of supporting indefinite binding extent is unclear, i would
suspect that dynamic extent would be the better alternative, but ask for clarification.

Received on Thursday, 6 August 1998 16:50:40 UTC