DataLink local_semantics optional column proposal

[Mark Taylor]

On Tue, 11 Oct 2022, Patrick Dowler wrote:

> This looks like a useful feature. Am I right in seeing this as a way to
> disambiguate where there are multiple links with the same {ID,semantics}?

Yes.  If {ID,semantics} was always unique, I wouldn't need this new column.

> If that's the case/intent, does it have to completely disambiguate? Does
> {ID,semantics,local_semantics} have to be unique? I'm not sure that's
> necessary and don't know what we would tell providers to do if they can't
> make the tuple unique...

It's desirable for my use case to disambiguate completely, but I don't
think there's any point in making it mandatory, so I'd say SHOULD.

I feel like since the data provider has a completely free hand in
assigning a local_semantics value, they should be able to provide
unique values if they want to ("thumbnail", "thumbnail_b",
"thumbnail_c", ...).  But since the column is going to be optional
anyway it's clearly a best-efforts value, and if they want to provide
an ambiguous local_semantics it's more use than not having one at all.

Mark

--
Mark Taylor  Astronomical Programmer  Physics, Bristol University, UK
m.b.taylor at bristol.ac.uk          http://www.star.bristol.ac.uk/~mbt/