[sc34wg3] The interpretation of facets

[Steve Pepper]

Michel,

At 20:31 26.04.2003 -0400, Michel Biezunski wrote:
>The issue you are raising about facets
>touches the question on what and how
>to reify and is one of the central issues
>the TMM is addressing.

No, the question of how to reify is very simple and it is handled in
the SAM. The Reference Model screws this up by conflating reification
with representation, as Lars Marius has shown in earlier postings.

>I urge you to
>read it again in this perspective. Maybe
>this time you will be able to understand
>it more than the first time you tried,
>because you would have an angle with
>which you can read it.

I already had this perspective when I spent two days studying N393:
It didn't help one iota. I am already convinced that N393 is a
complete f...ing mess that is of no use to anyone in its present state
(but also that it may contain insights that could prove valuable in
a new standard five years from now). I do not intend to waste time
re-reading it. I have more important things to do.

>I don't think that the SAM as exposed
>now is able to address this, because
>precisely of some of the weaknesses it
>has about reification.

Be specific. What weaknesses does the SAM have in this respect?
The proposal I have given for the interpretation of facets works
very nicely with the SAM as-is ... it even handles mnemonics,
as I have shown in my most recent posting. If you think the SAM
is defective in the area of reification, kindly tell us why.

>By the way these issues
>need to be studied at length. In
>the current state, we are facing
>serious interoperability problems
>such as how to interpret facets.

As you have seen from my postings, I *am* studying this at very
great length ;-) And I am providing concrete proposals to solve
the mess, not pointing people off in a totally irrelevant
direction.

>With the mails recently written about
>considering conformance as a secondary
>issue, there is not much to be left
>to make it a standard.

Who on earth has written mails recently about considering
conformance as a secondary issue? No-one I know of. All I've
seen is a very thoughtful suggestion from Lars Marius that
there are limits to how much conformance one can define to a
data model. Those limits are internal consistency. Most of
the conformance we need to define will be in terms of the
specifications that *use* the data model (XTM, HyTM, CXTM,
TMQL, TMCL, TMTL, etc.).

I don't call that "considering conformance as a secondary
issue." I would rather call it considering conformance as a
major issue, thinking deeply about it, and making serious
proposals.

Or were you referring to something I missed?

>I hope that the
>discussion in London will help making
>progress in that direction, because
>this is a serious concern. Why having
>a standard if we only can guess what
>it does?

Precisely. Up to now, with 13250 as it is, we have only been
guessing. We are very concerned about that. That is why
absolutely everything we have written in the SAM, the XTM spec,
the CXTM spec and the recent postings about HyTM has been
aimed at clarifying what 13250 is actually saying.

>A solution to that is to deserialize
>the SAM into a microscope
>such as the one provided by the TMM.

It is not a microscope. It is a Total Mega-Muddle. I believe
that "deserializing" the SAM in terms of N393 will be a huge
waste of effort. But if you think it's important, go ahead
and do it.

>It is a way to make explicit
>what's happening under the cover
>and stop relying on specific
>implementations to decide in our
>place. I am afraid that the problem
>which Martin is confronted to is
>only the tip of the iceberg. Let's
>make sure that the standard will work
>the way it is supposed to and can
>actually qualify as a standard.

I agree that 13250 barely qualifies as a standard today. What
we should be doing - the editors in particular - is clarifying
the *existing* standard, not inventing a *new one*. Let's get
13250 right before moving on to the Next Thing.

>My responsability as one of the editors
>of the standard is to make sure that
>we use the up-to-date tools we have
>in our disposal to handle the pending
>issues. You are asking me to go
>backward, I don't see why I should
>do that.

I'm not asking you to go backward. I'm asking you to do your
job as an editor and finish the thing you started. If you
really don't want to, that's fair enough. But then you should
say so and let someone else take over.

>Just for the record, although I was
>present when facets were removed, not
>in Dallas as you said but in Swindon UK
>if I recall properly, I was one of the
>few (or maybe the only one) who
>opposed to that decision.

My recollection of the date of the final decision is different,
but no matter. You give me even more reason to think of you as
Mr. Facets. Do you still wish that facets were in XTM? If you
do, why not argue the case. Martin will support you.

If you are concerned about the issue of facets, as you claim,
I look forward to your technical comments on my proposal.

Steve