[
Lists Home |
Date Index |
Thread Index
]
Jonathan Robie wrote:
> Yes, the problem with XML Schema is that is pseudo-formal. It has the
> precision of prose and the easy readability of mathematical notation. The
> Schema Formal Description tries to make this completely formal, but this
> work has never been completed.
To come back on this important issue, IMO a specification should use
either a formal description language or plain natural language (or both).
This reminds me of a specification language named SDL (normalized by the
CCITT) I have been using in the 80's. The great thing with such a
language was that you could run simulations and test your
specifications. I remember having specified a project using SDL and
having detected design flaws on tests scenarios before having written a
single line of code (it was not eXtreme Programming but eXtreme
Specifications!).
This language was not so far from what is needed to specify the
implementation of a schema language since it was especially well
featured for describing finite state machines.
I have not used it since the 90's, but there seem to have been some
convergences with UML-RT recently.
If the purpose of a spec such has W3C XML Schema part1 is to describe
the validation algorithm why not using such a language completed by a
plain English description?
Eric
--
See you in Paris.
http://www.afnet.fr/afnet/net200x/programme.html#T9
------------------------------------------------------------------------
Eric van der Vlist http://xmlfr.org http://dyomedea.com
http://xsltunit.org http://4xt.org http://examplotron.org
------------------------------------------------------------------------
|