[
Lists Home |
Date Index |
Thread Index
]
From: "Simon St.Laurent" <simonstl@simonstl.com>
> The more or less complete difficulty of
> representing the PSVI directly in XML suggests that there are some
> genuine efficiency issues to consider - whatever direction you may
> choose to follow them.
I suspect that almost all that is needed to represent the PSVI modestly is
something like
<a b="c">zzzz</a>
becomes
<a psvi:type="something" psvi:validation-attempted="full" psvi-valid="true" >
<psvi:attribute name="b" psvi:type="someothertype" >c</psvi:attribute
>zzzz</a>
In other words, the problem for a PSVI-in-XML is that attributes are
atomic. Rather than doing fancy naming tricks, it would be easier
to allow structured attributes, as part of XML 2.0, with the semantic
that in the XML infoset, those elements are attributes, just as a
namespace attribute in a document is not an attribute in the namespace
and an encoding PI is not a PI in the infoset and an xslt:attribute element
generates an attribute in its output.
Cheer
Rick Jelliffe
|
