[sc34wg3] Conformance

[Lars Marius Garshol]

* Jan Algermissen
|
| The following email discussion is taken from the threat "Questions
| on N0396: (8) Conformance":

Hey, Jan, are you sure you mean what you wrote there? :-)

* Lars Marius Garshol
|
| Depends how much is in the wrapper. If there is an equivalent
| structure underneath the C layer would be conformant, too. An RDBMS
| would *not* be conformant, even though you could build a SAM
| implementation on top of it.

* Jan Algermissen
|
| What I read from the above is this:
| 
| If the SAM will eventually have a conformance clause it will constrain
| the internal details of implementations. The SAM would not 'allow' for
| a conforming implementation to be based on an RDBMS.
| 
| Lars, is that correct?

No, absolutely not. An RDBMS would *not* be conformant alone, but the
SAM implementation on top of it + the RDBMS *would* be conformant.
 
| Without the intend to insult Lars (who has himself questioned the
| value of a conformance clause in the SAM) I claim that this is plain
| nonsense.

I agree.
 
| What is the reason to constrain the structural representation that
| implementations use internally? Given the entity relationship model,
| does that constrain the internals of the software of database
| products? **What** is the possible benefit for the SAM to require
| conformance to its conceptual model?

I have no idea.
 
| To me, conformance makes only sense at the API/query language level,
| since there it will enable interoperability between applications.

I think you should add syntax and schema language, but other than that
I agree.
 
| So, **what** is the purpose of a conceptual model for topic maps (be
| it N0393 or N0396)?

The SAM is *not* a conceptual model, it's the instrument we use for
defining the interpretation of the two syntaxes, the relationship
between them, and we will also use it to define a conformance testing
tool (CXTM), and finally it will be the foundation of the constraint
and query languages.

That's its purpose and sole reason for existing.

As for the purpose of N0393 I am still not clear on that. It appears
to intend to be a model for integration of information from different
data models, but there might be other reasons for it to exist as well
which I don't know about.
 
| Clearly, there are portions of both, N0393 and N0396 that do make
| sense to constrain applications (e.g. merging behaviour, value
| equality...), but what is the purpose of the conceptual data model?
| We need it, sure, but what for?

You lost me here, I'm afraid. What do you mean by "the conceptual data
model"?
 
| Sorry to be provocative, but I try to get everyone to think about
| this!

Thank you for doing so. I think this is crucially important, and would
love to hear more input on this.

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50                  <URL: http://www.garshol.priv.no >