Lists Home |
Date Index |
----- Original Message -----
From: "Sean McGrath" <firstname.lastname@example.org>
> [Marcus Carr]
> Perhaps I'm atypical, but most of my XML data is based on (if not
> >validated against) a DTD. If my DTD doesn't mesh with my data, it
> >means that my data is wrong. Reports about the death of DTDs continue to
> >be greatly exaggerated...
> I use DTDs for their most excellent powerful, intuitive, consise,
> extended regular expression notation.
powerful? Comparing to metamata's JavaCC ( the next step
after YACC ) DTDs are'nt powerful, because DTDs
power is comparable to YACC and JavaCC is 'better',
intuitive? Well ... this is subjective, but I don't find them
to be intuitive ... You mean emulating inherinatce
with macroprocessing is intuitive? Hmm...
regular extressions? Well, I'd say that with regular
expressions we have regexprs version 1 ( grep or
'see if there is a match') and regexprs version 2
( perl's $1, which is 'if there is a match - take this
and that part of matching string' ) DTDs are at
the level of version 1, so I think that maybe they're
kinda ... obsolete ... ( And JavaCC kinda 'got it right' ... )
> I eschew all the other stuff heaped in there from the SGML days
> where possible.
Such as? I'd really love to know about the best alternatives
to DTDs. Many thanks.
> It is truly exasperating to see *ML.org sites being created at a rate of
> knots and their drivers getting all "deeply committed" to W3C XML
> Schema without thinking through the implications.
> Especially for document-oriented XML applications.
Yes, Schema is a monster. But that does not mean that
DTDs are really good.
> I said it at the Orlando conference and I'll say it again here. If w3C
> is so complex that you *need* tools to visualize it then its XML 1.0
> base will not save us from the clutches of proprietary tool vendors.
No doubt. Schema is a monster. Perhaps, some tweaking of
DTDs may be better than building on XSD, but DTDs as-is?
> DTDs are not perfect, there are multiple alternatives, some better than
> others for document-oriented XML applications.
Well, I don't understand your point then. If there is something
better than DTDs, why are you defending the DTDs?
> I used to think that XML would finally break down what I always thought of
> an accidental divide between "data" and "documents".
But there *is* a divide. For example, for 'data' the order of properties
is actually not really important and for documents there is a
( I think sometimes misleading ) view that the order is important.
And also that mixed content stuff ...
> It does appear to me now that the chasm is too wide for XML to cross. Our
> will revisit this one mid-century.
30+ years ago Mr. Brooks told us that the data structures are the most
important thing in software development. Since then there was not too much
progress in this area, I think. I agree that it would take a very long time
to get some better structures rather than keep using arrays, variables and
hashtables. At least those assembly rudiments, called 'pointers', are now
optional. Not bad for 30 years.
PS. And of course, we all know that at least one of XML standards
(XSL FO) can not be expressed in DTDs, so ... There is some
problem, either with XSL FO ( I doubt that, because XSL FO
is driven by people with many years of printing experience,
so it should be that they know what they 'really need' for printing ),
or with DTDs ( more likely ).