[sc34wg3] Merging issues in SAM

Lars Marius Garshol sc34wg3@isotopicmaps.org
11 Oct 2002 20:20:41 +0200


* Bernard Vatant
|
| [SAM section 4.3]
|
| Should not it be:
|
| [...]
| 3. Apply merging to the [topics] property of A until no duplicates
| remain.
| 4. Apply merging to the [associations] property of A until no
| duplicates remain."

Indeed it should. Thanks for catching this! Now fixed.

This section may need to be revisited, however. I would prefer to
formulate the metamodel in such a way that it requires this merging to
be performed without this section having to say it explicitly. I think
that is both cleaner, less likely to lead to inconsistencies and
problems, and also more in line with how we would want this to
actually work.

Unfortunately that has implications for the conformance requirements,
so I am not sure what the right solution is yet. Input is certainly
welcome. 
 
| 4.6  Merging association items
| 
| Is this section conformant to XTM 1.0 Annex F.2.4 Association
| equality principle ?

Yes. Note the second sentence of the third list item:

  "B's values are equal to those of A, and need therefore not be
  taken into account."

Note that the SAM counterpart of F.2.4 is really the last para of
section 3.8.
 
| As a matter of fact, F.2.4 makes pretty obvious sense to me, whereas
| SAM 4.6 is quite opaque:
| 
| "Two association items, A and B, *may* be merged by following the
| procedure below ..."
| 
| Under which necessary/sufficient conditions? Does that mean they
| never *have* to? Do I miss something?
| 
| Thanks to enlighten me ...

This is a good point, and it does identify something that is very
clearly a weakness of the current document. This ties back to the
metamodel comment I made above: the SAM should make it clearer under
what circumstances merging is to be performed.

This is an issue for which no localized fix may exist, so it will take
a revision of the SAM, and some thinking, to fix this. I've made a
note of this complaint, which we've gotten from many other directions
as well, so we'll give it a good try. Suggestions are of course more
than welcome.

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
ISO SC34/WG3, OASIS GeoLang TC        <URL: http://www.garshol.priv.no >