[sc34wg3] Aug 3 meeting: Issue (term-tm-processor), Issue (xtm-implicit-constraints):

[Lars Marius Garshol]

* Lars Marius Garshol
|
| TMCL is about specifying application-specific constraints, whereas
| the SAM constraints are constraints that must apply to all topic
| maps because of how topic maps work.

* Robert Barta
| 
| .....whereby this boundary might be somewhat artifical, while
| useful.

Artificial? It's certainly a matter of non-trivial judgment to decide
where the boundary should lie. If that's what you mean by artificial I
agree.

| A halfway decent TMCL will have to impose structural constraints as
| well.  Better to have it in SAM, though.

We'll have it both places. SAM will specify some invariant constraints
that apply to all topic maps as part of the model itself, such as "all
variant names must be contained in a topic name", "all association
roles must have both a role type and a role player".

TMCL will allow additional constraints such as "every composer must
have a name in the unconstrained scope", "every composer must have a
born-in association with a topic of type city", and so on.

Both are structural constraints, but some are ontology-specific, while
others are not. The former go in TMCL, the latter in the SAM.

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