[
Lists Home |
Date Index |
Thread Index
]
On Fri, 2003-08-01 at 10:38, Simon St.Laurent wrote:
> mc@xegesis.org (Mike Champion) writes:
> >There are some increasingly important corner cases where all this is
> >needed except that there is NOT that excess capacity that XML
> >text/markup processing requires. People are trying to figure how to
> >have most of the XML cake and eat it too: Keep Infosets, XPath, DOM,
> >XSLT, XQuery, validation, etc.; give up some interoperability, but get
> >back some performance. You may well be right that XML is ultimately
> >destined to be more trouble than it's worth in such scenarios, but it
> >will take a lot of failed experiments to prove that. Don't hold your
> >breath.
>
> I didn't think it would happen within breath-holding time. The
> experiments have barely begun. Even if it doesn't happen at the
> meeting, I hope it happens in practice, over time. Give binary formats
> some momentum of their own again and see what happens.
>
>
the biggest single speed up would come from doing what mime did - have
an optional length specified - which if present must be correct. then if
a parser found a length/size whatever attribute (still in ascii) it
could quickly find the end tag. if the attribute is missing it parses as
it does today.
of course that might mean venturing into the dangerous world of
"reserved attributes" which is not in the spirit of xml 1.0, but then
neither are binary formats and this would at least preserve the human
readability of xml.
rick
|
