VocInVO2: Desise change, License

[Markus Demleitner]

Hi all,

On Wed, May 13, 2020 at 12:25:50PM +0100, Mark Taylor wrote:
> On Tue, 12 May 2020, Markus Demleitner wrote:
> > > So, the proposal is to make this more explicit in this way:
> > > 
> > >   "terms": {
> > >     "EQUATORIAL": {
> > >       "label": "Equatorial",
> > >       "description": "Umbrella term of all equatorial frame.  Only use for old, pre-FK4 equatorial coordinates."
> > >     },
> > 
> > This is now how the desise in the RDF repo looks like.  This has the
> 
> Thanks Markus for that change.
> 
> However, now I've got that much I want more...
> 
> The current (revised as above) WD has top-level entries including
> "terms", "deprecated_terms", "preliminary_terms" and "wider_terms";
> the example looks like this:
> 
[...]
>      "deprecated_terms": [
>        "B1875.0"
>      ],
>      "preliminary_terms": [
>        "ICRS2"
>      ],
>      "wider_terms": {
>        "ICRS": [
>          "EQUATORIAL"
>        ],
>        "ICRS2": [
>          "EQUATORIAL"
>        ]
>      }
>    }
> 
> I'd like to suggest instead collapsing the deprecated, preliminary
> and wider information into the "terms" object, so that it looks
> instead like:
> 
[...]
>        "ICRS": {
>          "label": "ICRS",
>          "description": "As defined by 1998AJ....116..516M.",
>          "wider": [ "EQUATORIAL" ]
>        },
>        "B1875.0": {
>          "label": "Bonner Durchmusterung System",
>          "description": "Deprecated term for the reference system implied by BD/CD",
>          "deprecated": ""
>         },
> 
> To me that looks like a more logical and readable way to organise
> the information.  The convention for the flag-like attributes
> "preliminary" and "description" is that the flag is considered
> asserted if its key is present in the term, regardless of the
> corresponding value.
> A "use_instead" attribute could be added for deprecated terms
> as well if required.
> 
> Anybody want to agree or disagree that this would be an improvement?

Since no immediate cheers or boos came up:  Well, once the values of
the terms dictionary are dicts themselves, I have to admit Mark's
proposal is rather natural.

The reason I had done it differently so far is that I was mimicking
revovo's API, which had different functions (well, methods) for the
different use cases ("find deprecated terms", etc) -- which, looking
back, was perhaps not the greatest of designs in the first place.

Following Mark's proposal doesn't make the use of the data structure
much harder (for instance, term in voc["deprecated"] becomes
"deprecated" in voc["terms"][term] in Python), and given I tend to
lean towards Mark's claims about "logical", I think I'll put it into
the document some time next week unless there's more dissent,
together with a clarification that desise is intended as a
VO-specific, document-typed API to the RDF rather than some special
RDF distribution format.

I'd then issue another WD with updated desise documentation (and a
few other changes) towards Friday or so.  It would of course be great
if I could then already include additional feedback, so if anyone who
reads this could spare half an hour to skim the document until then,
I'd be very grateful:
http://ivoa.net/documents/Vocabularies/20200326/.

Thanks,

         Markus