[
Lists Home |
Date Index |
Thread Index
]
At 4:59 PM -0400 4/12/04, Bob Wyman wrote:
>Elliot Rusty Harold wrote:
>> DOM and SAX at least can handle documents that do
>> not have infosets, and I think XPath/XSLT can too
> What does it mean to say that an XML document "does not have
>an infoset"? I'm a bit puzzled by the words here. I thought that any
>XML document could be described via an Infoset. What am I missing? (My
>apologies if this is a silly question or if I've missed something very
>fundamental in the XML specs...)
They're at least two such cases that exist, as John Cowan pointed out before:
1. Documents that are namespace malformed
2. Namespace well-formed documents that use relative namespace URIs
I've encountered both of these in practice. There may be other cases,
but these (or more specifically case 1) were what I was thinking of
here.
The problem in the reverse direction is much larger though. There are
many, many infosets that do not correspond to well-formed XML 1.0
documents. This has been a major problem for various specifications
based on the Infoset including XInclude. By starting with the
infoset, rather than XML itself, technologies tend to lose track of
some critical rules like "element names may not contain white space"
or "an element may not have two attributes with the same name." These
rules are not enforced in the infoset, only in real XML.
The mapping between infosets and XML documents is neither 1-1 nor
onto, even when lexical issues like white space inside tags is
ignored.
Remember, despite what people keep saying the infoset is *NOT* a data
model for XML. It is *NOT* a replacement for the XPath data model,
the DOM data model, or any other data model.
--
Elliotte Rusty Harold
elharo@metalab.unc.edu
Effective XML (Addison-Wesley, 2003)
http://www.cafeconleche.org/books/effectivexml
http://www.amazon.com/exec/obidos/ISBN%3D0321150406/ref%3Dnosim/cafeaulaitA
|
