First feedback on the GitHub usage

[François Bonnarel]

Hi Ole,all,


Le 03/12/2019 à 17:00, Ole Streicher a écrit :
> Hi all,
>
> On 02.12.19 22:36, Patrick Dowler wrote:
>> On Fri, 29 Nov 2019 at 09:45, François Bonnarel wrote:
>>> So, I'm not really positive about moving to github in these conditions.
>> I guess all I can say here is that I agree with everything you said and
>> maybe the only "issue" is that github was seen/expected/presented/sold
>> as the bug magic replacement for everything that was going to make it
>> all better and we'd be able to produce new standards in a few months. Of
>> course, it doesn't do that at all -- we all still have to do the real
>> hard work and github can help with a portion of it. From my point of
>> view, it helps with (i) track things we agreed to do or change, (ii)
>> accept implementation of those changes, (iii) review the implemented
>> changes. It is not far off to be able to (iv) automate testing that
>> changes did not break the document build, (v) automate publish the
>> latest cutting edge document where everyone can see it*, and (vi)
>> automate publishing to the IVOA doc repo under some condition (maybe
>> something like tagging the repo as WD-Foo-1.1-20191225).
> One addition to point (i) here: Github issues and pull requests are also
> well-suited to discuss and agree on changes and extensions of the
> standard. One of the goals of the Github transition was to make this
> process more transparent and reproducible.
I strongly disagree there. This is what was the decision of the TCG in 
the face to face meeting before Groningen interop
>
>   * Document process in github -- discuss here - review and more open
>     discussion in Sunday AM session
>       o proposal: WG chair and vice chair are admins of the repo; they
>         give document editors read-write permission -- then only
>         chair(s) and editors can merge pull requests
>       o proposal: a team/group made up of TCG (all? chair & vie-chair)
>         will have permission to create new repositories
>       o we do need to assign a license to each repo (proposal:
>         http://creativecommons.org/licenses/by-sa/4.0/-- need exec
>         approval and something should end up inside the docs)
>       o /Summary from discussion:/
>      o
>           + /TCG has agreed that we will move the development of
>             standards documents from volute (kindly hosted by GAVO) to
>             github/
>           + /TCG chair/vice-chair will manage permissions and create
>             repositories for standards (1 repository per document)/
>           + /WG chair/vice-chair will admin and manage permissions on
>             a set of repositories (~10)/
>           + /WG chair/vice-chair and specified document editors will
>             have write permission (to merge changes)/
>           + /details to be demonstrated at Sunday AM session/
>           + /TCG recommends that all documents be licensed under the
>             Creative Commons Attribution ShareAlike
>             <https://wiki.ivoa.net/twiki/bin/edit/IVOA/ShareAlike?topicparent=IVOA.IvoaExecMeetingFM87;nowysiwyg=0>(http://creativecommons.org/licenses/by-sa/4.0/)
>             license: DocStd
>             <https://wiki.ivoa.net/twiki/bin/edit/IVOA/DocStd?topicparent=IVOA.IvoaExecMeetingFM87;nowysiwyg=0>change/
>
>           + /*ACTION: *PD/JE: Exec review decision when stable of the
>             doc license choice/
>
This looks like a practical  managment enhancment procedure and NOT like 
a fully new specification document philosophy.

Regarding the authorship of specifications and the discussion among 
authors :

Specifications are not "open code", they don't have to be managed the 
same, they are more like scientific papers.

  Will authors of scientific papers in general be happy to expose their 
discussions openly from scratch ?

The discussion in Working group, TCG, votes by TCG and Exec are similar 
to a review process.

What is proposed below is making the authors role disappear. Only editor 
and IVOA(overall?) community is left.

I don't think we have to go to this direction.

regards
François

>   So, the idea is in my opinion
> *not* to agree on a change behind closed doors, but to use the Github
> tools (issues, PRs) to find an agreement, to enable others to follow and
> contribute to this process, and to have a public log of it for later
> reference.
>
> There is no need to "restrict" some part of the development process to
> the authors. On contrary, the "magic" of Github is exactly that
> transparency. And this will pay off in the long term.
>
> Best regards
>
> Ole

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/interop/attachments/20191204/32258342/attachment.html>