pvdbosch
commented
5 months ago From what I've learned about BEST identifiers, I wonder if we really need to describe the different parts of the identifier separately?
Each addressable object could just be identified by a single globally unique URI (string) that's internally composed of namespace+object+version. This could be compliant to ICEG standards and simplify representation by having a single string instead of a triplet of parameters.
If we expose the identifier's components, we risk mixing business data with identification, which is not a good practice (e.g. birth date and gender in SSIN). Because regions handle versioning differently, mutations should be modeled explicitly in a data model and exposed through APIs instead of implicitly from identifiers.
Aside from this discussion:
- do we want to mandate number format for parts of the identifier? If not, I'd just talk about "identifier" instead of "number".
- "which enables a user to distinguish it from the neighbor addresses," => "... distinguish it from other Belgian addresses" (not only the neighboring ones)
2024-04-22 Mail Eddy Corthouts (PM project BEST):
Voor enkele definities in fedvoc zou het BeSt team volgende aanpassingen willen voorstellen:
1/ namespace
Weglaten eerste zin. In de 2e zin “object” toevoegen tussen “same” en “identifier”, dus “same object identifier”
2/ Object identifier
Huidige definitie
Voorstel aangepaste definitie: “a number uniquely identifying the object within the namespace”
3/ Class Adress
Huidige definitie
Voorstel aangepaste definitie:
“An address is a spatial object that in a human readable way identifies a fixed location of a property (building, parcel, …). For this purpose, an address has an identifier, e.g. an address number, which enables a user to distinguish it from the neighbor addresses, as well as a geographic position, which enables an application to locate the address spatially.”