Lists Home |
Date Index |
On Wednesday 26 February 2003 19:18, Cavnar-Johnson, John wrote:
> As fascinating as this is, what does it have to do with subsetting XML?
It rose from discussion of whether XML could have a meaningful common data
model at all, or if it was only globally a synctic constraint on character
strings; my argument is that if it is a syntactic constraint on character
strings at all (meaning, XML can be defined by "the set of all possible
strings which are well formed XML documents") then, due to the nature of the
constraints that define that set (well formedness), it's inherently a tree
Even if you don't *have* to base your processing around the tree structure,
as SAX doesn't, the tree structure is still there. Likewise, somebody using
DOM bases their work entirely around the tree structure and ignores the fact
that it might once have been a linear stream of something.
Then somebody said it was a DAG, which I said it isn't, but then two people
came up with challenges to this; one saying that he can avoid re-specifying
duplicated subtrees in instances by putting them into entities and
referencing them, and another by saying that you can define an otherwise
non-special type of leaf node to represent a pointer to another part of the
tree and then state that the resulting relationship is supposed to represent
identity, but the first is missing the point and the second is just decribing
a way you could encode a DAG into a tree.
But the overall theme of the subthread is my argument that failing to
formally specify a data model for XML doesn't prevent there being an inherent
tree structure due to the structure of the definition of XML, but it doesn't
make a DAG :-)
Does that about cover it?
A city is like a large, complex, rabbit