[
Lists Home |
Date Index |
Thread Index
]
On Tuesday 18 February 2003 02:44 pm, C. M. Sperberg-McQueen wrote:
> I'm astonished to see document people disowning the idea of
> datatyping, as if better datatyping (and the failure of #PCDATA
> to provide the constraints we wanted, for things like numbers
> or dates or other simple datatypes) had not been one of the
> most frequently mentioned desiderata in discussions of markup
> among professional DTD designers, in the years 1986-1998.
Then again, not all of the people dealing with documents in the years
1986-1998, cared that much about DTD's to that level of detail.
A fairly large number of people dealing with SGML documents, including myself,
regard validation as but one of the useful things you can do to an SGML
document, and maybe not even the most useful one. That's one reason for
well-formed XML.
Documents have a lifetime, and in my world, are often dynamic entities that
pass through phases, many of which would be valid against no DTD I care about
(let alone schema!). The decision of marking something up as
<date>1999-01-01</date>
or
<date calendar="gregorian">
<year>1999</year>
<month>01</month>
<month>01</month>
</date>
is only relevant in the context of a given application, which may or may not
care about validation and strict typing. Typing is like everything else: a
compromise between purity and practicality (or perhaps generality).
|