[sc34wg3] Conformance

Lars Marius Garshol sc34wg3@isotopicmaps.org
27 Apr 2003 21:14:44 +0200

* Jan Algermissen
| I am only concerned with the abstract information structure (which
| != SAM).  So, are you saying that this struture is "the instrument
| we use for defining the interpretation..."?
Well, in the current XTM, HyTM, and CXTM drafts that instrument *is*
the SAM. If you try to make the point that "in theory this could be
something other than the SAM" then obviously that is true.
| I am talking about the data model only, not relational calculus nor
| relational algebra. Suppose there was only Chen's original paper, no
| more - what's the purpose the data model fulfills?

I don't know. That's hypothetical, and I don't think it really has
much bearing on our situation.
| Anyhow, if the abstract information structure of N0396 (or N0393) is
| not constraining the application internals, what does it mean to
| define merging rules in terms of that structure?
| It seems to me we are saying something like: "A conforming
| application must appear (to the 'user') as if it actually
| implemented the abstract information structure."  But, duh- what
| does that *mean*?

It means that given (XTM-doc-1, TMQL-query-2) conforming
implementations must provide a particular result where they behave as
if they have performed certain merging operations before applying the
query. Ditto for TMCL-schema-3.

Also, when testing the implementation with CXTM the implementation
must produce CXTM output where the right information items have been

I would also argue that a user of some engine with a non-standard API
could point to XTM 1.1 + SAM and say "look, you don't implement
merging as it says here, that must be a bug", but that user is
obviously expressing a desire to have the API behave in conformance
with a certain view of how the API maps to the SAM, which is probably
appropriate, but not something with which the standard should concern
itself, in my view.
| Also, N0396 says:
| "Merging is a process applied to topic maps in order to reduce the
| number of redundant information items representing the same
| information."
| What is the meaning of this, if we are not interested in the
| internals?

The point is that when you connect the serialization syntax with some
other specification through SAM you get the expected behaviour. How
that is achieved is of course outside the realm of standardization.

| (I absolutely don't care how many objects the Omnigator uses for a
| single subject internally, all I want is a certain behaviour 'to the
| outside')

I agree with that.
| So, what is it that we are constraining at the N0396 or N0393 level?

In my opinion, nothing. The constraining all happens in the other

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