Lists Home |
Date Index |
On Fri, 28 Feb 2003 16:23:41 -0500, Norman Walsh <firstname.lastname@example.org> wrote:
> But I wonder how rock-solid the basis actually is. That's the part
> that bothers me. If you said "XML Kernel" is XML 1.1 w/o DTDs would
> someone come along a week later and say they must have a subset that
> doesn't have PIs in it? Or comments? Or attributes? Or #PCDATA? I dunno.
> It seems unlikely, it seems like the task of parsing those things
> isn't very onerous. But we said that about 1.0 with DTDs, didn't we?
I meant "rock solid" in the sense that applications that depend only on
elements, attributes, and text are extremely likely to be interoperable, or
portable, or reliable, or whatever one uses to measure "solidity." It's
quite true that there's not rock-solid *definition* of the "core" or
"kernel" ... e.g. the de-facto SOAP subset excludes PIs and DOCTYPE
declarations but XML-SW includes them. And that's before we start talking
about namespaces, at which point the former allies start to turn on one
another :-) It would be very interesting to analyze the costs and benefits
of drawing the line one one side or the other.