[
Lists Home |
Date Index |
Thread Index
]
"Alaric B. Snell" wrote:
> Any volunteers, in the meantime? :-) I'm pro-binxml, so will need an
> anti-binxml viewpoint to balance...
My 'anti-binxml viewpoint':
XML as UnicodeWithAngleBrackets is two operations of process short of BinaryXML.
I take it as given that this is by design; it is certainly the foundation of
XML's greatest usefulness and the essence of interoperability via XML. As I said
in this permathread two and a half years ago
(http://lists.xml.org/archives/xml-dev/200007/msg00945.html), the binary impulse
is a 'retreat from the radical and most useful implications' of well-formedness.
In order to derive a binary instantiation from an instance of syntactic XML
requires two operations: first a parse and then the construction of some
particular data structure upon the output of that parse. Syntactic XML requires
that the creator and every consumer of a document agree a priori on that parse:
the specification of a general-purpose parser inherent in the XML 1.0
Recommendation is the necessary substance of that contract. Beyond that parse,
however, nothing essential to XML either requires nor defines the nature of the
process by which a tree, an infoset or any other data structure is built on the
output of parsing. Most specifically, neither the content model which might be
expressed in a DTD nor any other schematic is binding upon any consumer of an XML
document.
The simple essence of XML is its parseability in conformance with the Rec. The
binary camp begins by obviating that necessary proof of any instance's nature as
XML. The first premise of the the binary camp is that there is a canonical data
structure or infoset of which every XML instance is a document-order-serialized
realization. That obviates all of the infinite other structures which might be
constructed on a particular occasion upon the output of a particular parse of a
specific syntactic instance. Far from being wave and particle or the right and
left hands of darkness and light, these two visions are mutually exclusive,
utterly incommensurate, and diametrically opposed in both their intent and in the
outcomes which they necessarily effect.
XML is authoritatively specified as syntax. What each consumer might do with a
particular instance of that syntax is not specified, and as we joyously discover
in practice, is effectively unlimited. Any other specification of 'XML', of the
processes by which it is expected to be processed, or of the data structures into
which it is to be instantiated precludes from the first the vast majority of
those possibilities.
Respectfully,
Walter Perry
|
