TMCL requirements; Was: [sc34wg3] a new name for the RM

[Mason, James David (MXM)]

To follow up what both Mary and Martin say:

We're talking about two audiences here: implementers and users. (At the
moment, those audiences overlap heavily, but if our standards succeed, they
will incresingly become separate, as has happened with SGML/XML.)

My view is that we write standards primarily for implementers; we write TRs
primarily for users. 

If the RM is about a base-level definition of what TMs are, then it sounds
like a standard. It ought to be, as Martin says, "human parsable" (I find
the current 13250 pretty opaque). It ought to go from graph theory, or some
other suitable starting points, to the point at which we can build the SAM.

If we want to say what all this means for people building ontologies, etc.,
then that ought to be a separate document, a TR. I strongly support having a
TR, maybe more than one TR, to help users like me who can't make it through
the thorny stuff!

We do need to guard, however, against the RM becoming too theoretical. Back
in the old days of the ODA/SGML War, I used to needle the ODA guys because
they didn't define anything with a production grammar (like SGML), BNF, or
any other quasi-mathematical formalism. Eentually they decided to outdo the
SGML folks by starting a "Formal Definition of ODA". It turned out to be a
great tree-killer, volume after volume of symbolic-logic notation that
amounted to an existence proof for ODA. Absolutely no use to anyone, either
users or implementers. It kept them occupied. It also kept them from doing
any useful work, and that helped lead to the cancellation of their project.
Mission accomplished!

Jim

-----Original Message-----
From: Martin Bryan [mailto:mtbryan@sgml.u-net.com]
Sent: Monday, January 27, 2003 3:00 AM
To: sc34wg3@isotopicmaps.org
Subject: Re: TMCL requirements; Was: [sc34wg3] a new name for the RM



In response to

> >* Michel Biezunski
> >|
> >| Remember the standard is not only aimed at implementers, it's also
> >| for information users.
>
Mary wrote:

> I really disagree with this statement. These standards are not for
> information users. We do need documents for them too, but the RM, SAM,
> TMCL, TMQL are not for them. We do need a good tutorial-like introduction
> though and something equivalent to Tom Bray's annotation of the XML
> standard (note: this was not part of the standard, but possibly read by
> 1000X more people than the original standard itself) The standard was read
> by the parser writers (I hope :))

The point that Michel was making is that TMCL statements, whatever form they
might take, need to be human parsable so that users can validate whether or
not the constraints have been met. If the language is so convoluted that
only a program can implement it then it will be a complete failure. Unless
the end-users understand what they are doing, and why, writing any formal
descripition of the language is irrelevant. Whether the tutorial forms an
introduction to the standard, and informative annex or a separate TR is
irrelevant. What is vital is that the human understandable version is
available for review at the same time as the formal material so that we can
get adequate feedback from the user community as well as the few developers
able to understand any formal definitions.

Martin Bryan

_______________________________________________
sc34wg3 mailing list
sc34wg3@isotopicmaps.org
http://www.isotopicmaps.org/mailman/listinfo/sc34wg3