From W3C Wiki
Jump to: navigation, search

Social Web Working Group Teleconference

09 May 2017

See also: IRC log


tantek, cwebber, ajordan, sandro, aaronpk, eprodrom
cwebber, eprodrom


<ajordan> present_

<cwebber> I can scribe

<cwebber> scribenick: cwebber

<ajordan> cwebber++

<Loqi> cwebber has 15 karma


tantek: ok, let's review and approve last week's minutes


<tantek> +1

<eprodrom> +1

<sandro> +1

<ajordan> +1

<aaronpk> +1

RESOLUTION: approved

tantek: next item: let's talk about the next telcon
... we had proposals to do talks either every week or every two weeks
... it made sense previously to do talks every week as opposed to every two weeks

sandro: I'm thinking until we're caught up and run out of things to talk about on meetings we should probably be meeting every week

eprodrom: I just wanted to say I wonder to what extent this is dependent on the schedule for primarily ActivityPub, to some extent WebSub
... I think we've talked a little bit about extending the group if possible, urgency possibly goes up or down depending on whether we do that

tantek: I agree, particularly in getting activitypub and websub in terms of checkmarks in terms of CR -> PR
... to put it in a positive way, as long as we make normative changes we need to issue new CRs for or are updating the test suite, then we probably need to meet every week
... is that similar to what you're saying?

eprodrom: yes that's exactly it
... I'd hate to miss a deadline because we missed a meeting

tantek: it sounds like the chairs and staff contact are in pretty strong agreement. if nobody objects I think we can make a call

aaronpk: I'm good with next week

<ajordan> cwebber: that was me

<aaronpk> haha i wasn't even dialed in yet

<Loqi> hahahaha

tantek: and part of this is we need to set expectations that we'll do weekly calls as long as there's no objections

<eprodrom> +1

<ajordan> +1

<aaronpk> +1 i'll be in UTC+2 next week tho

tantek: this isn't that big of a surprise anyway, we're basically moving back to our normal weekly telcon schedule until swe're confident with at least ActivityPub (and WebSub?) moving to PR


<eprodrom> Yes

<eprodrom> All good

RESOLUTION: we'll keep meeting weekly as long as there's work to do on CR->PR on websub / AP

tantek: I believe AS2 and MicroPub end PR period on Thursday
... so Amy and Sandro, any other news on PR vote on AS2 & Micropub?
... and AS2

sandro: Amy is still on her offline hiatus

<sandro> you can't hear me?

<eprodrom> I can hear everyone

<sandro> tantek we hear you


<sandro> but you cant hear us!

<eprodrom> I can hear Tantek and Aaron and Sandro

<eprodrom> I feel like this is a logic problem

<aaronpk> this is like my ham radio calls

sandro: summary is nothing has changed

tantek: in that case, I will reiterate my encouragement to all of you to reach out to your AC reps for all of you
... and encourage them to at least support the specifications if not commit to implementing them
... the more positive notes we can get the better. but at least we have no objections
... I'm postponing Post Type Discovery


tantek: where are we with the test suite


aaronpk: the test suite is very nearly done, subscriber tests are there
... a nice list of all the test cases there
... that's based on some of the github discussions that have happened. fortunately it doesn't require any normative changes in the spec because it's just behavior that shoudl happen in terms of http
... I'm in the middle of finishing the hubtests. the two main ones are done
... a basic subscription request, and there is a test for the signature (?)
... some hubs are public, some are part of publishing software like mastodon, wordpress... so I basically had to make two different flows for starting those tests
... yeah, those two are in place, I had to work on the edge cases
... the one thing I didn't do in the tests in order to speed things up is I didn't do anything to make it store the results
... I didn't do anything to have it store the results, so it's just a tool that helps you fill out the implementation report


tantek: ok so that begs the question, is there an implementation report template that an implementer can fill out manually based on the results of the test suite they get

aaronpk: yep there's three of them, one for each role... the subscriber template is complete, and once I finish the hub test I'll have that
... I wanted to avoid previous problem where someone filled out report without tests available
... we have one for subscribers, will do for publishers

tantek: the sooner we can put out a call for people to fill it out the better

aaronpk: as soon as I do edge cases for publishers I can do taht

tantek: if we're ready for subscribers, we can ask that now

aaronpk: all I need to do is write the publisher report template

tantek: great, then we can get the pub and the sub of pubsubhubbub
... when do you think this test suite for hubs will be completed

aaronpk: I'm hoping before next week, before the end of this week, it'll be done

tantek: that includes implementation reports as well?

aaronpk: yep

tantek: sounds good, do we know of implementations that will pass them, or?
... are we waiting more on "we have to find more implementations to document, encourage more implementing"

aaronpk: I've been testing with a few as I"ve been going, superfeedr already passes it, and mastodon is very close
... I haven't tested as many subscribers because I don't know as many to test

tantek: do you know of at least two implementations of subscribers?

aaronpk: superfeedr is a subscriber but I haven't run it through tests as a subscriber yet
... there's woodwind also, but I haven't seen the results of that yet

tantek: ok

<sandro> Curious if passes as a publisher (+hub?). It claims to implement PuSH.

tantek: any new issues needing group discussion for websub?

aaronpk: don't believe so

tantek: any editorial changes you need to publish as an updated CR?

aaronpk: not at the moment

sandro: we do have 4 uncategorized open issues

aaronpk: 97 is a discussion one intended to remain open until PR, don't know how to tag that
... I think 98 could be resolved with some editorial language, 99 could be resolved as soon as implementation reports are in place, 102 is a question I had when I was implementing test suite, might not require any normative changes, wanted julian to weigh in


<Loqi> [aaronpk] #102 Why should subscribers return 2xx on invalid signatures?

aaronpk: 102 is when a hub delivers a notification that includes an incorrect signature, this is when it's expecting it to sign a notification
... when a hub returns an incorrect signature, it still returns 200, but I was wondering why not 400 bad request but I don't have any way to test that without a different status code to read
... ben pointed out that allowing the receiver to return 200 here allows processing here
... that makes sense to me, do others have thoughts on other reasons to return 200?

sandro: I don't see ben's reason as a reason to say MUST, seems like a MAY. Only MUST I can think of is if you think it's a security risk. Given this is a human keyspace not a computer keyspace, I don'tt think guessing is a threat

aaronpk: the other thing I could see is the hub is going to treat a non-200 status code as an error
... this may deactivate a subscription
... this may be one reason, but at the point that a hub is incorrectly ??? a signature, I'm not sure that's a valid case it would get into

sandro: you'd want it to stop in this case I'd think. My inclination is this should be a MAY
... I agree this should at least be editorial

aaronpk: yeah I think it needed some explaination regardless of outcome, but I wanted Julian to chime in because he probably knew where it came from

tantek: could we have you at least propose some non-normative explanatory language?

aaronpk: not until I know more about why this was in here

tantek: I mean more the point about async from ben and sandro's point about reasoning for a MAY

<Loqi> Tantekelik made 1 edit to Socialwg/2017-05-09

aaronpk: yeah I'd rather wait

tantek: ok we'll look again next week
... could you ping julian again, aaronpk ?

aaronpk: yep... I am going to mark it as editorial since it will be some explaination

sandro: and if we change it to a MAY...
... it's hard to see that as requiring a new CR because it's loosening conformance criteria

tantek: it's non-normative because it's loosening conformance requirements
... I'd like to resolve it one way or another next week
... in particular aaronpk I'd like you to figure out what you'd like to figure out what you'd like to do with this issue by next week

aaronpk: ok


<Loqi> [aaronpk] Upon further thought, the safest thing to do is the second approach I described.

<Loqi> * The old topic URL should send a redirect to the new topic URL. This will provide a seamless transition for any clients polling the topic URL.

<Loqi> * When the subscript...

aaronpk: ok I'd also like to quickly talk about issue 98. I think we can get a resolution quickly
... I realized if you followed HTTP you'd handle subscriptions automatically
... doesn't require any special handling on the subscriber which is nice
... I would like to add an informative section describing this situation so there's some guidance in the spec on how to migrate

tantek: ok
... seems reasonable, what do people think

sandro: agreed

<eprodrom> +1

tantek: I think esp since it's editorial we can leave it to editor's discretion

aaronpk: ok
... will write it up

tantek: write that up then why don't we accumulate any other editorial changes, if we're going to do a new CR with editorial changes let's do it next week instead of today

aaronpk: ok
... we can publish new CR without restarting the process?

tantek: yes for editorial only changes

sandro: we did that with ActivityPub like 3 weeks ago

tantek: feelf ree to edit the editorial draft as usual

aaronpk: that covers everything

tantek: goal for next week is to resolve that issue and handle the CR

<eprodrom> I can scribe

<eprodrom> scribenick: eprodrom

<tantek> scribenick: eprodrom


cwebber: NextCloud released a beta with AP support
... leaving OAuth out of the test suite
... Not going to refactor, just going forward with current design
... moved forward with CR2
... 9 May 2017 CR


<Loqi> [sandhawke] #221 Controlling availability to search

cwebber: A search engine was getting the public feeds on Mastodon and had some responses


This is what audience targeting is for


<tantek> or what an extension would be for

<sandro> "As I said above, I don't think a flag is the right way to address this issue."

<sandro> ***

No, this is explicitly what audience targeting is for

<tantek> I see it as different but am listening

<tantek> eprodrom: are you able to minute this?

<Zakim> sandro, you wanted to clarify issue

ajordan: did you check the Nextcloud implementation?

cwebber: no, only saw mention

ajordan: there seems to be a desire for fine-grained control, public-but-not-searchable

sandro: question of flags not being sufficient
... the thread of 50-60 posts is worth reading through

<sandro> that thread is

<Loqi> [Antoine Aflalo] Want a full text #search engine for toots? Try: made by @vhf

sandro: we will have CG discussions on systems besides SW specs

cwebber: fine with extensions

tantek: let's use extension routes for doing this
... should not need to block AP

eprodrom: we could use more well-known addresses for fine-grained public control (like PublicNotSearchable)


<Loqi> [annando] #196 How to differentiate between posts and private (direct) messages?

<tantek> ack

cwebber: we have deferred this without a solution

eprodrom: I think we have a solution and the commenter is refusing to understand it

tantek: we can add editorial guidance, and move on from there

sandro: if we have more time, let's not close it

tantek: where are we with the test suite?


cwebber, do you want to go back to scribing?

<cwebber> scribenick: cwebber

<eprodrom> cwebber, thanks sir

sandro: as we talked about it last week, we talked about an extension as well as a rechartering. Extension is to take vastly increased enthusiasm and mastodon momentum in activitypub
... twenty-two new user interfaces in the last month and half a million users seems to be plenty of reason to take more time
... and I'm fairly confident w3c will agree

tantek: the other observation last week is the release of last week and their use of micropub and social apis
... their whole area has been to do replies using webmention and vouch, so that was the other piece that would go into this extension or rechartering, but seems like another area
... we have a spec and there's interest

sandro: do you have numbers to show that to be more than one person's project?

aaronpk: there's a lot of kickstarter backers

<aaronpk> 3080 backers

tantek: there's multiple implementations and the stuff

<aaronpk> and lots of people building new micropub clients to talk to

tantek: there's at least 3k people with accounts... not half a million, but that's a big jump

<aaronpk> anecdotal numbers since i've been watching their slack

sandro: yep the kickstarter is great. I'll add that, I hadn't seen it

tantek: that's an argument for Vouch IMO
... this is a momentum argument

<ajordan> we're at the hour

tantek: the other one to consider which has a weaker case is salmentions
... the only reason to bring up is the SWAT0 which we've strongly agreed on
... I'd like to try to get it in scope if we're doing a rechartering for that reason
... we have multiple implementations using it

eprodrom: yeah to me what I'd like to see is that we do as limited a rechartering as possible... take the things we already have going that may or may not be ready to close and get them closed. for me and... I know cwebber does not agree.. but I think splitting up the mature part of activitypub of c2s go out and the s2s part which we're not sure we can test make that a second half
... split it into two things, see AP get finished
... I think that might be a good way to suggest getting things done
... we've already extended once, I worry we might not be able to get the whole part done

<ajordan> can whoever's chair ask if we can extend by 10 minutes?

<eprodrom> scribenick: eprodrom

<sandro> +1 <ajordan> can whoever's chair ask if we can extend by 10 minutes?

cwebber: We have a simple way to do testing of s2s auth
... we had the CG meeting last week, lot of enthusiasm
... CG group wants to get AP finished
... so that GNU Social and Mastodon gets caught up


<cwebber> scribenick: cwebber

<ajordan> +1

+1 to extending

<sandro> +1 extending 15min

<aaronpk> +1

tantek: ok we're extended to 9:15

<Zakim> sandro, you wanted to clarify extension vs recharter

sandro: to clarify the process, extending is what we did in December, and what I'm suggesting we do now
... it's basically a w3c management decision saying this group is doing ok, and it doesn't involve moving to the advisory committee or change scope. it might involve agreement that it goes to this narrow thing, but it's relatively straightforward
... my expectation is that if things are picking up the way things are going, we might want to recharter, which might mean some people come and go, but a rechartering makes sense (it would mean 20 members get on board)
... but those are very separate things between rechartering and extension
... if we get the extension to weave mastodon into the fold, would it make sense to weave in vouch and salmentions?

tantek: the other way to frame those if you want is iterating on webmention
... these are the extensions we've iterated on since the rec and I think that's grey area enough to ask what your comfort level is

sandro: I will put that on the agenda

tantek: the sort-of-obvious thing is I haven't had time to do updates on Post Type Discovery

sandro: I don't think we'd need to say a "limited extension", we'd leave our hands not particularly tied
... at which case we have a continuing chartter

tantek: I want to address eprodrom, he was addressing potential c2s and s2s separation
... while I don't disagree on maturity, from extension or rechartering it makes sense to keep both
... I would rather not tie our hands
... as long as it doesn't put a risk of us not being able to put on them

sandro: I don't think so

<eprodrom> I don't think we need to

tantek: any idea on time?

sandro: I feel like 6 months is the best compromise. we could maybe ask for 3 months

<eprodrom> I like to ship

tantek: I think that's wise to increase the runway

<sandro> +1 shipping!

+1 on shipping also :)

<eprodrom> cwebber, is there an issue for that?

eprodrom, there is

<eprodrom> I'm happy to talk to that problem

cwebber: mastodon very specifically wanted clarity on webfinger

tantek: we want to avoid starting a new fight on that though

ajordan: should we ask for 3-6 months?

<Zakim> tantek, you wanted to note something about maintenance

sandro: best to not be wishy washy and just clear about what we want

tantek: the other point raised to me in other capacity is increasing desire from w3c management for working groups to do maintenance of their specs including folding in erratta etc
... esp if you spec is being close to being done etc
... normative fixes etc
... so you could see a .rev etc
... this is something to consider for everything that we're doing
... from my understanding the w3c management *wants* to hear on how we are doing maintenance

sandro: seems to me pretty clear given our CG plan that we have a pretty good plan there
... I'm assuming aaron and chris are willing to maintain the specs they've edited

tantek: first half is getting eratta edited and done in there, second half is getting them published as official w3c docs with IP protection, etc
... my understanding is they'd like to see that in a WG

sandro: when we get to that, yes

<ajordan> eprodrom: is the Webfinger issue from a couple minutes ago

<Loqi> [cwebber] #194 Include informative section suggesting how WebFinger users can migrate towards ActivityPub adoption?

<Zakim> cwebber, you wanted to just mention next CG call before this call ends

<aaronpk> swicg pronounced "sink"

PROPOSED: Extend for 6 months to incorporate increased developer feedback

<ajordan> +1

<sandro> +1


<aaronpk> +1

<eprodrom> +1

<tantek> +1

RESOLUTION: Extend for 6 months to incorporate increased developer feedback

tantek: in that case I want to plant a seed in peoples' heads for face to face possibilities
... maybe we can talk next week
... in particular put down on your calendars the TPAC meeting


tantek: see you all next tuesday!

<eprodrom> Thanks tantek !

<eprodrom> Thanks cwebber for scribing!

<tantek> cwebber++ for scribing!

<Loqi> cwebber has 16 karma

<ajordan> cwebber++ (again)

<Loqi> cwebber has 17 karma

<tantek> trackbot, end meeting

Summary of Action Items

Summary of Resolutions

  1. approved
  2. we'll keep meeting weekly as long as there's work to do on CR->PR on websub / AP
  3. Extend for 6 months to incorporate increased developer feedback

[End of minutes]