New WD of Vocabularies in the VO 2.0

[Markus Demleitner]

Dear Semantics WG,

Yesterday, I've uploaded the second working draft for Vocabularies
in the VO 2; it's now in the document repo at

http://ivoa.net/documents/Vocabularies/20200326/

Here's a list of the major changes, roughly in order of perceived (by
me) importance:

(1) The "Usage without RDF tooling" no longer uses RDF/XML but
instead the ad-hoc JSON schema desise I've mentioned last week:
http://mail.ivoa.net/pipermail/semantics/2020-March/002677.html

I'd particularly like feedback on this.  I have to admit that the
more I get used to the idea, the more I like it.  I'm just not sure
whether that's a good or a bad sign.

(2) Lex UAT.  In the context of this work I realised that I was
rather ignora^[bC bold when I wrote "Use the UAT" in VOResource 1.1;
some of the problems are discussed in
http://ivoa.net/documents/Vocabularies/20200326/WD-Vocabularies-2.0-20200326.html#tth_sEc2.2.12.
Long story short: I'm now rather convinced we need a UAT mirror at
IVOA's RDF repo, with clear rules how we do this both RDF-correctly
and so our "dead simple" clients can still work with it.  An attempt
to deal with this is the new section 5.3, externally managed
vocabularies.  This is regulation done rather to get out of a bad fix
than something I'd choose to do in a fresh design.  But I couldn't
work out anything smarter.

(3) I'm now requiring a link to something using the proposed term in
VEP addition requests.  This means that trying to introduce terms
ahead of time is now formally forbidden outside of initial vocabulary
creation.  *I* think that's what we should do, but if you disagree,
this would be a good moment to discuss this.

(4) I've added a non-normative soapbox section on how people should
(IMNSNO, that is) go about defining new terms.  This goes just a few
steps into formal semantics, but in a maximally naive way.  If you
think that's not a good use of spec pages, let me know and I'll throw
the stuff out again.

(5) There's a bit more language on VEP management: It's now
explicit that it's the chair of the Semantics WG who uploads VEPs and
assigns the running number (where I've tried to make clear that at
that stage that's a purely formal activity), and I've put in a
mechanism with which related VEPs reference one another.

(6) Finally, I'm saying how I think IVOA vocabularies should and
should not be referenced.


As usual, I'm grateful for any kind of feedback.  If there's no
devastating criticism, I'd probably push this to PR in, perhaps, June
or July, at least if there's a client or two already doing something
with desise (I'd say datalink, VOTable or VOResource validators would
be obvious first adopters; I'd *really* like to see a datalink client
exploit the vocabulary, of course, and I personally might teach
DaCHS's ADQL engine a gavo_voc_expand(vocname, term) function or so).

So, any help is as much appreciated as is feedback.

Oh, and if anyone involved with trying out and/or getting straight
the first WD (in particular, participating in the VEPs and providing
feedback) would like to get on the author list, just let me know.

Thanks,

         Markus