[
Lists Home |
Date Index |
Thread Index
]
--- Elliotte Rusty Harold <elharo@metalab.unc.edu>
wrote:
> 3rd Option: Do nothing. Leave XML as it is. It
> works. It's proven.
> It's not perfect, but it's damn good, and hits a
> really nice middle
> point between the needs of many different
> communities. Don't trim it.
> Don't create alternate syntaxes. Just use it.
Uhh, it works ... sortof. Pheonomenally great for
proof of concept work, but the people who push its
limits complain like hell. It sortof meets the needs
of different communities, but numerous devils infest
the details.
As for "just use it," everyone here probably knows
perfectly well how to use it, but we find it a pain in
the butt to type (hence Don's Son of SML proposal),
program against (hence all sorts of API innovations,
including yours), use in high-performance environments
(hence one stand of the "binary XML" thread), use in
limited bandwidth environments (hence the other
strand), write schemas for (hence RELAX NG), etc.
It's good, but not "damn good." Like everything else
in the world, the stuff that works well is being
copied, and the stuff that doesn't work well is being
tweaked. This (and all other manifestations of this
permathread) are about how to evolve the body of XML
theory, practice, and standardization in such a way as
to preserve the good and gracefully deprecate the
cruft. There *is* a danger of throwing out the
interoperability baby with the crufty bathwater, I'll
freely admit. I will assert (based on the history of
programming languages, DBMS systems, operating
systems, etc.) that if XML does not evolve quickly to
meet the concerns of the power users, it will simply
be tossed aside in the next great Paradigm Shift....
baby, bathwater, tub, and all.
|
