search

muze-nl / jsontag

Typed JSON
MIT License
8 stars 2 forks source link

add more unique identifier fields? #6

Open steltenpower opened 2 years ago

steltenpower commented 2 years ago

phone is already an example (if no abbreviated numbers are used, but only those starting with "+", country code, and the rest), but many more are around:

IBAN (bank account) EAN ISBN-13 (book) DOI (scientific output) ORCID (researcher ID) citizen id (BSN in the Netherlands) passport ID IPnr EMEI and many more

All with their own value constraints e.g. ISBN has a check digit

poef commented 2 years ago

As mentioned in #5 I'd like to keep the base set of types limited. But I can see that I haven't created a generic ID type, which may be useful. I could imagine a type like this:

<id class="IBAN">"NO 93 8601 1117947"

The problem here is that it is easy to become confused about the term 'id'. It has many different meanings. There is an id attribute in the spec already, but that is not the same as an id property. This id type would add another level of confusion.

Adding seperate types for each of these is, I think, a bad idea. This would overload the specification and make it much more complex to implement a parser.

For some of these examples, it might be better to use the type address, e.g:

<address class="IP IPv4">"192.168.1.1"

But I do agree that the set of base types is not set in stone, I'd like to get some real world tests done and then evaluate the current selection.