First feedback on the GitHub usage

[Markus Demleitner]

Dear VO community,

Although I'm not sure interop@ is the right forum for this
discussion, I'd like to chime in on one point that I think is
actually rather central and has been guiding me in promoting version
control for spec authoring in the first place:

On Wed, Dec 04, 2019 at 03:58:21PM +0100, François Bonnarel wrote:
> Specifications are not "open code", they don't have to be managed the same,
> they are more like scientific papers.

I don't think that's true.  I am rather convinced that specifications
are, essentially, abstract programmes.  Well: they should be, and
there are formal languages that make that explicit.  We don't use
these formal languages in the VO for better or worse, but even though
it's natural language we write, a good and useful specification
prescribes sets of behaviours quite as precisely and reproducably as
a piece of code.

Where it fails to do that, interoperabiltiy and implementability
suffer -- and we have seen our share of examples for that in the
history of the VO.  Viewed from that angle, the requirements for
interoperating implementations and validators are attempts to
encourage debugging specifications much like unit (≈validators) and
regression (≈implementations) tests help debugging programmes.
Whereever we take these requirements seriously, our specifications
have significantly fewer, well, bugs.

So, there is a lot to be said for developing specifications with
even more care, review, and validation than programmes, and that
makes it very desirable to re-use the tools developed for that.

           -- Markus

(who, truth be told, still isn't happy that we're caving in to
platform pressure and will probably adopt proprietary tech; but
that's a different question, and since I acknowledge that socially,
a move to github will fairly certainly benefit the IVOA, I won't
speak out against it)