Mike Champion wrote:
> FWIW, I completely agree (e.g. with Sean) that this is seldom a
> bottleneck in applications, but it does seem to be a bottleneck in some
> industrial- strength infrastructure implementations. That's a major
> reason given by the XMLP people in pushing-back on calls for SOAP to
> support the full XML spec.
Not that anybody on any occasion has any introduced any statistically
significant data generated under controlled circumstances. It's just a
bunch of anecdotal hand-waving at the moment. BTW, I don't disbelieve
that this could be a significant optimization; but if these engineers
were working for me, and walked into my office asking for non-conformant
subsetting and didn't have some quantitative data to back up their
suspicions, they'd be bouncing out of the office pretty damn quick. -Tim
