Re: Clark's commentary

/ Tim Bray <tbray@textuality.com> was heard to say:
| James also specifically suggested simply leaving all of the DTD
| machinery.  The degree to which this would simplify the specification
| work is almost beyond belief.  I think I could sit down and write
| an XML 2.0 draft (1.0 - DTDs + namespaces + infoset) in about 3
| days elapsed.  -Tim

I think it's too early to do that. We don't have good replacements for
entity declarations yet (IMHO) so we need to keep some DTD
declarations around.

If I was working on it, I think I'd try to restructure the spec so
that "DTD validation" was a black box with respect to the rest of the
spec. Then one could imagine writing *a different document* that
described how to replace that black box with a different black box (or
with none at all for the simply well-formed case).

The operative word in the preceding paragraph is "try". Without really
trying to do it, I'm hesitant to predict how successful the effort
would be.

I was going to write

  To bring this back around to the topic at hand, web architecture, I
  think this is a good example of how a problem might have been broken
  down into smaller problems: validation as separate from well-formed
  parsing

but alas it's clearly not that simple since entity references require
both parts. Ah, well, I still think it's a good example in spirit: we
should be writing small, discrete, composable specs to the largest
extent possible.

                                        Be seeing you,
                                          norm

-- 
Norman.Walsh@Sun.COM   | A man can believe a considerable deal of
XML Standards Engineer | rubbish, and yet go about his daily work in a
XML Technology Center  | rational and cheerful manner.--Norman Douglas
Sun Microsystems, Inc. |

Received on Monday, 7 January 2002 14:01:12 UTC