Re: Split off ActivityPub CG or WG


On 14/10/2023 11:50, Melvin Carvalho wrote:
>
>
> pá 13. 10. 2023 v 16:15 odesílatel Pierre-Antoine Champin 
> <pierre-antoine@w3.org> napsal:
>
>
>     On 13/10/2023 10:31, Melvin Carvalho wrote:
>>
>>
>>     út 10. 10. 2023 v 9:22 odesílatel Pierre-Antoine Champin
>>     <pierre-antoine@w3.org> napsal:
>>
>>         Dear all,
>>
>>         may be a middle ground would be to
>>
>>         1- keep the SocialWeb CG as one umbrella group with a wide
>>         scope, and
>>         2- create one (or several) more focused WGs
>>
>>         This is, for example, what happens with the Credentials CG
>>         <https://www.w3.org/community/credentials/>,
>>         that has spun off the Verifiable Credentials
>>         <https://www.w3.org/2017/vc/WG/> and the Decentralized
>>         Identifiers <https://www.w3.org/2019/did-wg/> WGs.
>>
>>         I also wanted to raise awareness about the notion of
>>         "maintenance WG":
>>         a WG that is chartered to produce no new recommendation, but
>>         only to maintain existing ones:
>>         applying errata, making editorial and substantial
>>         improvements (but not adding brand new features).
>>
>>         Would it make sense to start by creating an Activity Pub
>>         Maintenance WG for maintaining AS2 and AP?
>>         Nothing would prevent this group from requesting a
>>         rechartering at some point to start working on ActivityPub2,
>>         whenever they and the CG find it is timely.
>>
>>         Nothing would prevent, also, other WGs to emerge from the CG.
>>         The CG would remain the place where the broad community
>>         incubates and exchange new ideas. WGs, on the other hand,
>>         could focus on delivering specs.
>>
>>
>>     Hi Pierre
>>
>>     I thought about this for quite a bit more.  The question arises
>>     on how to define "maintenance".
>
>     The term"maintenance group" is not explicitly mentioned in the W3C
>     process, but we use this term to designate a group that has no new
>     deliverable in its charter. Per the process, it means that all it
>     can do is apply class 1 to class 3 changes [1] to those documents
>     (anything but brand new features).
>
>     [1] https://www.w3.org/2023/Process-20230612/#correction-classes

>
>     Note that this also include updating (or creating) a test suite
>     for the spec.
>
>
> This makes a lot of sense.  Changes of class 1-3 + test-suites.
>
> The question is who determines what is class 1, 2 or 3. A spec change 
> could be inadvertently put in the wrong bucket.
At different stages of its lifecycle, A REC-track document undergoes 
wide review. This is were this kind of problem would be caught.
>
> However, there is no workflow right now, which ensures that changes 
> will not inadvertently be put in the wrong bucket.
>
> If an open source effort does this in parallel, then it's possible to 
> compare notes, and see how it compares with the test suites.
>
> It will take a while for the test suite to have full coverage.  
> There's the NLNet one and also evan's old one at as2.rocks.  Plus a 
> few more.
>
> Perhaps it's an idea to wait and see which test suite achieves the 
> most coverage in the next 6 months or so.
>
> My slight question is what happens if test suites disagree?
>
>       best
>
>>     Perhaps the first step is to fix all of the editorial errors
>>     (e.g. clear syntax errors, grammatical mistakes) in the docs.
>>
>>     Some of the so-called "errata" are opinions on whether something
>>     should go one way or another, rather than,  It's often the
>>     opinions of 1-2 people on a given issue, without clear reasoning,
>>     or real world stats to back it up.
>>
>>     The goal of a maintenance update should be to provide a more
>>     robust interoperable space:
>>
>>     https://en.wikipedia.org/wiki/Robustness_principle

>>
>>     However, there is interop and there is also "bug for bug"
>>     compatibility.  We run the risk of introducing new bugs that get
>>     institutionalized as technical debt.  It's hard at this point to
>>     categorize what's a bug and what's a feature. We dont have
>>     criteria to decide that right now, and we need to.
>>
>>     Granted the open source maintenance effort is not much butte, but
>>     I think an important step will be to improve tooling and
>>     processes.  For example the test suite should be helpful.  And it
>>     should figure out whether we have true interop or just bug for
>>     bug compatibility, and give the network an opportunity to fix
>>     bugs and heal, in a backwards compatible way.
>>
>>     The process of correcting simple mistakes from opinionated errata
>>     should be separate and we should do one before the other.  I
>>     think that's already how things work?
>>
>>     I think a proper robust maintenance update needs to include the
>>     whole ecosystem, not just a few people making wiki edits. 
>>     Although Evan's primer is going to be a very useful input, we
>>     need more docs like this, and ability to compare notes.  And it's
>>     going to take 6 months to 1 year to get the tooling in place,
>>     namely a test-suite. And possibly a FEP for the maintenance
>>     update in parallel.
>>
>>     We can probably all agree that the tooling we have for this job
>>     will be better than we have now, and so will the documentation. 
>>     I think a maintenance WG is a good idea, but we shouldnt dive
>>     into it until we are sure we are upgrading the network in a
>>     backwards-compatible, testable, and bug-free way.  In preparation
>>     for this we can document existing practices ( my effort is
>>     started here: https://socialdocs.org/ ) in detail, and also
>>     codify them in the test suite.
>>
>>     That could give everyone what they want, Evan writes the primer,
>>     the whole community go through and compare it with what's in
>>     production, open source makes a maintenance FEP, and cross checks
>>     things with a test-suite to see how that matches with what's in
>>     production.  When we reach a point where the docs and the
>>     tooling, and the proposed maintenance items are agreed upon by
>>     the whole ecosystem, that's probably a good time to charter a WG
>>     and update to perhaps AP 1.1 and AS 2.1.
>>
>>     Thoughts?
>>
>>         my 2¢
>>
>>         On 04/10/2023 20:06, Christine Lemmer-Webber wrote:
>>>         BTW, I agree with Evan: the CG *is* the home of ActivityPub currently.
>>>
>>>         I made the suggestion to make a separate CG or WG for ActivityPub is
>>>         mainly to provide more focus within the standards process, and it was
>>>         primarily a suggestion about if a WG was made in particular.  It could
>>>         be that this is a *sub-group* or outgrowth of the SocialCG, but I am
>>>         also okay with it not happening.  I appreciate the massive amount of
>>>         work Evan is doing organizing and reviving the CG right now, by the way.
>>>
>>>         But I see the "official home" of the specification, and the authority of
>>>         what constitutes ActivityPub, as the SocialCG at present.  I think
>>>         SocialHub is an excellent place to coordinate, and works with the
>>>         SocialCG.  But the SocialCG is where AP's editing authority resides, and
>>>         I think that's good and appropriate and it was designed to be that way.
>>>
>>>           - Christine
>>>
>>>         Evan Prodromou<evan@prodromou.name>  <mailto:evan@prodromou.name>  writes:
>>>
>>>>         SocialHub is a great place for CG members to chat.
>>>>
>>>>         I think the FEP process is a great way to develop extensions.
>>>>
>>>>         The core ActivityPub and Activity Streams 2.0 specs are maintained by this CG.
>>>>
>>>>         Evan
>>>>
>>>>         On Oct 4, 2023 03:21, hellekin<hellekin@cepheide.org>  <mailto:hellekin@cepheide.org>  wrote:
>>>>
>>>>           We do have an ActivityPub Community Group -- a Special Interest Group
>>>>           actually, in the form of the SocialHub.
>>>>
>>>>           If Aaron Parecki thinks it's good to keep the SocialCG and work with
>>>>           ActivityPub within a broader context of the Social Web protocols, then I
>>>>           see no reason to split again. We can continue ActivityPub ground work on
>>>>           the SocialHub, relay to the SocialCG and get the best of both worlds.
>>>>
>>>>           The SocialHub was created to give momentum to the ActivityPub community
>>>>           following the ActivityPub Conference held in Prague in 2019, and
>>>>           organized very generously by Sebastian Lasse. It was a great success and
>>>>           we anticipated much work to do that would become much noise for the
>>>>           SocialCG mailing-list, since this list was larger than just ActivityPub.
>>>>
>>>>           If now the people we wanted to avoid spamming are fine with getting the
>>>>           heat, I see no reason to move away and apart. On the contrary, I feel
>>>>           like we are in a situation where we have a real grassroots community
>>>>           that is grounded in free software and works on Codeberg and the
>>>>           SocialHub, and a standards-oriented community group who can relay and
>>>>           give body to already chewed on ground work. This is the best situation
>>>>           we can imagine, where the grassroots implementors lead the way and the
>>>>           standards-oriented people renders that body of work normative.
>>>>
>>>>           I am not a driving force in the specification process, so I'm happy
>>>>           whatever decision is made, but I want to underline both the grassroots
>>>>           effort that have been going on over the last four years around the
>>>>           SocialHub, as well as the renewed interests by the Chair to consolidate
>>>>           the normative form of ActivityPub and ActivityStreams. This is a great
>>>>           opportunity to engage more people with more confidence in the process,
>>>>           and not isolate other protocols that, if they are less visible, are no
>>>>           less important to our common success.
>>>>
>>>>           ==
>>>>           hk
>>

Received on Tuesday, 17 October 2023 12:32:03 UTC