Re: Binary XML (was: Re: Draft minutes of 15 March 2005 Telcon)

> Why does this problem justify a binary XML standard?

This is precisely the question the TAG should focus on IMHO.  There is
little doubt that there are many real-world problems with XML's size and
parsing overhead. I don't think there's a whole lot of point in
disputing the Binary Characterization assessment of the problems.  Maybe
they will go away with better parsers and compression schemes, or maybe
Moore's Law will make them irrelevant, but none of those sound like safe
bets.  What people have done is develop significantly faster and/or
smaller encodings that have been shown to work for specific usage
scenarios in tightly coupled environments, e.g. between an API and a
DBMS engine, or for SOAP messages between two nodes built with the same
technology.  

What is MUCH less clear to me (and, as you might have noticed at the
Plenary sessions, several of my colleagues!) is whether a) there is
evidence that a single binary XML standard could optimize both speed and
size across a wide range of use cases, and b) whether that degree of
improvement justifies the very real social and economic costs that
another XML encoding standard would impose. Yes, there are benefits
(especially to the wireless industry), but could those benefits be
achieved with industry-specific standards that don't attempt to come up
with a global compromise? We all know from bitter experience that such
compromises are technically difficult to define and politically
difficult to justify. 

I would suggest that the burden of proof is on anyone writing a charter
for a followon binary XML WG to gather evidence to make the case that a
single binary XML standard could work across a wide range of XML
scenarios. Presumably there are enough prototype formats/implementations
out there to be analyzed according to the Binary XML Characterization
WG's criteria so that this should not be an unreasonably high hurdle to
overcome.   Then it could be the task of the TAG and AC to weigh
whatever those demonstrable benefits might be against the likely costs,
and use that information to rigorously justify a decision to pursue (or
not) a binary XML Recommendation-track WG.

Received on Friday, 18 March 2005 02:23:41 UTC