Re: GitHub Backup

On 25 Sep 2014, at 09:59 , Nick Stenning <nick@whiteink.com> wrote:

> On Thu, Sep 25, 2014, at 05:49, Ivan Herman wrote:
>> 
>> On 25 Sep 2014, at 24:47 , Randall Leeds <randall@bleeds.info> wrote:
>> 
>>> Noticed this from the minutes (sorry I wasn't there today):
>>> 
>>> | <fjh> we should explore automating pull of github for w3c archive as backup mechanism
>>> 
>> [...]
>> it's probably completely unnecessary from a disaster-recovery standpoint as long as any of 
>> us are actively involved in the project.
>> 
>> I agree. I do not think we have a real danger as long as some people
>> regularly update their local repository. [...]
> 
> Sorry, I perhaps made my point in a slightly roundabout way on the call
> yesterday.
> 
> GitHub is a great collaboration tool, and git is an excellent DVCS. I
> have absolutely no objections to our using either of these tools as part
> of our work. Indeed, I'd go so far as to say I can't recommend anything
> better!
> 
> That said, I'm always made a little nervous when commercial services
> like GitHub become so ubiquitous that they begin to be treated like
> pieces of public infrastructure.
> 
> As long as we're clear that our GitHub repository is solely a
> collaboration tool for an active working group, and not a canonical
> archive of any outputs of the WG (which, on reflection, is almost
> certainly how it was intended anyway) I'll shut up and stop causing
> trouble.

:-)

Maybe it is a good time to make clear what the role of github is in our process. TL;DR: github is indeed _not_ the canonical archive...

Official W3C documents (Working drafts, Recommendations, Notes, etc.) must go through an official publication route. This involves, for example, a collective decision of the group on publishing, but also styling requirements, requirements on the content like abstract, status section, headers, that sort of things. Once the document is 'ready', it is published on the W3C Web site, more exactly under http://www.w3.org/TR/. For example, 

http://www.w3.org/TR/2014/WD-tabular-data-model-20140710/

is a working draft of another spec-in-the-making. (The details of the whole publication process will come up when we get to that point, I let us not go into the details at this moment.) There is an expectation that a WG has regular 'heartbeats', ie, that it regularly publishes new versions of a draft, say, every 4-5 months (or even more frequently if it can). Documents in /TR are stored for eternity; if a new version of the same document comes out, older versions are kept in place. This also means that, on /TR, the whole history of a document's hearbeats are stored and never removed. E.g., an older version of the document above is at:

http://www.w3.org/TR/2014/WD-tabular-data-model-20140327/

However, documents need to evolve with everybody involved, subject to frequent editing, etc. This is where mercurial, CVS, or indeed github come in: it stores what we call, in our jargon, an editor's draft. The role of github is merely to store and provide a collaborative area for development for those editors' drafts prior to an official publication. E.g., the editor's draft for the document above is:

http://w3c.github.io/csvw/syntax/

Editor's draft can change at any time, can be removed, renamed, whatever, whereas official drafts on /TR are frozen. Editor's draft should not be considered as official by any means; documents on /TR have a stable status. (Actually, documents on /TR, while being Working Drafts, are semi-official because, well, they are drafts; they then become CR-s, ie, Candidate Recommendations, then PR-s, ie, Proposed Recommendations to be voted on by W3C members, and finally REC-s, ie, Recommendation, meaning the final and frozen specifications. All at the same place and with history archived.)

I hope this makes this a bit clearer. Does this answer to your concerns?

Ivan

P.S. B.t.w., Nick, I do not have your github account name yet...

> 
> Best,
> Nick
> 


----
Ivan Herman, W3C 
Digital Publishing Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
GPG: 0x343F1A3D
WebID: http://www.ivan-herman.net/foaf#me