[
Lists Home |
Date Index |
Thread Index
]
On Sun, 23 Feb 2003 09:54:39 -0800, Tim Bray <tbray@textuality.com> wrote:
> Etc... there's a word for these situations: "bugs". Bugs are a fact of
> life and we fix them and deal with them. When someone writes a
> specification for a potentially-very-important software library and
> writes the bug into the definition, that's a different level of
> seriousness.
It seems to me that Sean raises a more profound issue. If XML has brought
real interoperability benefits despite widespread non-conformance to the
spec (whether we call the non-conformance "bugs" or "features"), then the
argument that the benefits are threatened by non-conformant implementations
seems highly suspect. True, the conformance issues h's talking about are
much more at the level of nitpicks rather than major amputations such as
we're talking about here, but on the other hand I've never gotten the
impression from this list that DTD validation is widely used in data-style
and simple document applications such as one would expect to write for
J2ME.
I'd like to see a dialogue here, and to be honest I don't think calling the
J2ME proposal a "pseudo-XML botch" is a good opening. Labels aside (isn't
J2ME more or less implementing the XML-SW proposal? <grin>) Java users
benefit if the J2ME implementation is at least minimally compatible with
the great majority of XML data in actual use by prospective users; and the
W3C benefits if it uses the empirical evidence here to guide the
conformance level / profiles / subsets discussion. I think both could
learn from each other's needs and expertise here. One could argue that Sun
should wait until something like XML-SW is standardized as XML 2.0 (or a
conformance profile, or whatever), but standards that describe actual
practice will be more successful than those which prescribe ideal pratice.
|