[
Lists Home |
Date Index |
Thread Index
]
John Cowan wrote:
> Paul Prescod scripsit:
>
>
>>I hope I'm missing something: if I have ten XPointers in a document and
>>they each use 3 namespaces, do I necessarily have 30 xmlns declarations?
>
>
> No, you have 3 uses of the xmlns() scheme in each of your ten XPointers.
Each one "declaring" a namespace. ;)
> The xmlns declarations in the embedding document are irrelevant; after all,
I was asking about declarations using the xmlns scheme.
> that document need not be XML at all: it could be plain text, HTML,
> PDF, or hard copy.
IMHO, it is not wise to dumb down XPointer-in-XML to the same level as
XPointer-in-hard copy just to be consistent.
==
I am think that we need a scoped short-form mechanism to replace
entities. The more we put URIs as component identifiers in attribute
values, the more we need short-forms. If there is a growing concensus
(is there?) that QNames are not the way then maybe we need a declaration
mechanism specific to data. Perhaps it could be used both to declare
short-names for URIs and also for character entities.
Paul Prescod
|