Re: DTD Fragments and XML

Terry Allen wrote:
> A universal typology of human knowledge would do that for you, but
> there will never be one.  All you can do is refer to well known
> external semantics, e.g., via fixed attributes.  This is not an
> XML issue; the problem is the same in SGML.

Question: What does <P> mean in each context?

Typical SGML Document
=====================

<!DOCTYPE TEI.2 PUBLIC "-//TEI//DTD TEI Lite 1.0 //EN">
<TEI.2>
...
<Q>...</Q>

Typical XML Document (my prediction)
====================
#2. What does <P> mean:

<?XML version="1.0"?>
<Q>...</Q>

Answer:

SGML: "Q: contains a quotation or apparent quotation --- a
representation of speech or thought marked as being quoted from someone
else (whether in fact quoted or not); in narrative, the words are
usually those of a character or speaker;..." (from TEI specs)

XML: ??? Where do I go to find out ???

Allowing a sub-valid level of "correctness" without providing an easy
way of specifying namespaces is a hand-written invitation for people to
invent tags without describing them or providing a pointer to the
description. The actual consequences are more practical than reading the
documentation. How do you render "Q" in Braille? In the first case you
download the DTD with ICADD mappings, or the "Braille stylesheet" from
the "Braille Stylesheet Repository." In the second case you don't know
WHAT to download. How do you index them? In the first case, you look up
the mapping from TEI to whatever your standard DTD is. In the second
case, you can't know the mapping. etc. etc.

Here is my nightmare, and what I expect will happen. Someone will ask on
comp.text.xml, "How do I add a <FOO> element to my DocBook-XML document"
and they will get twenty answers back: "Just delete that useless DOCTYPE
line, use your element and everything will just work." In one minute
they will render their document useless for anything other than raw web
display using their own stylesheet. Something is wrong when we provide
such a powerful incentive for removing the namespace labelling mechanism
on their document and provide no alternate.

 Paul Prescod

Received on Saturday, 3 May 1997 16:36:39 UTC