From W3C Wiki

Social Web Working Group 5th face-to-face

16 Mar 2016


See also: IRC log


tantek, wilkie, dmitriz, eprodrom, rhiaro, aaronpk, shevski, ben_thatmustbeme, cwebber, tsyesika, sandro, Karli, AnnBass
ben_thatmustbeme, Ben Roberts, cwebber2, rhiaro, wilkie, sandro

Summary of Action Items

Summary of Resolutions

  1. reserve Thursday/Friday for meeting at TPAC in Lisbon
  2. Drop Action Handlers Editor's Draft from our list of working group drafts, without prejudice
  3. publish new AS2 working drafts with outstanding (agreed, reviewed) edits completed
  4. Resolve to publish webmention, micropub and activitypub pending changes agreed by the wg this face-to-face

Agenda Items

<tantek> trackbot, this meeting spans midnight

<trackbot> Sorry, tantek, I don't understand 'trackbot, this meeting spans midnight'. Please refer to <> for help.

<tantek> scribe: ben_thatmustbeme

<aaronpk> good morning!

<scribe> scribenick: ben_thatmustbeme

<scribe> scribe: Ben Roberts

<tsyesika> tsyesika+

tantek: lets do introductions

ben_thatmustbeme (I won't scribe intros)

tantek: the first thing we're going to be doing is editing the agenda

<Karli> ok

tantek: we have demos and soem admin items to schedule

<aaronpk> scribenick: aaronpk

<shevski> hello

<shevski> i can't find you guyes

<wilkie> the room?

<shevski> yep, the room for the f2f

<wilkie> stata center, 4th floor common

<shevski> i did a fair amount of wandering around the 4th floor..

<rhiaro> G449

<rhiaro> shevski

<shevski> okay, on the 5th floor now. will come and have another look

sandro: i took an action item to propose a list of labels for organizing issues, i have that to present now
... probably before we go through issues lists

<ben_thatmust> scribenick: ben_thatmust

tantek: the next goals after that will be to looking at what is needed next for taking drafts to CR
... lets move work on publish updated working drafts to tomorrow

eprodrom: lets move demo of validator to before taking AS2 to CR discussion

tantek: this fully schedules what we had for goals and items to schedule thus far.
... do we have any other items we'd like to add?

shevski: what about reviewing implementations?

tantek: (schedules it) NOW i think we are done
... anything else?

AnnBass: do we have anyone on talky?

tantek: no, not yet

<eprodrom> Arnaud: are you joining by Talky?

tantek: the next telcon is the 29th, no telcon next week

<eprodrom> ben_thatmust++

tantek: mostly because most of us are busy here next week

<Loqi> ben_thatmust has 8 karma

tantek: the next F2F is already scheduled for portland
... the weekend before the meeting is the IndieWebCamp summit too
... right now we have 5 RSVPs to the next F2F


tantek: please add yourself one way or another if you can go or not
... that will help aaronpk with venue planning etc
... the next F2F after that is in september in Lisbon Portugal as part of TPAC
... by a quick show of hands how many can make it to that?
... that was about 50% so it should be enough of a critical mass
... we need to submit for space
... i proposed thursday - friday that week
... if anyone else has any WG they are a part of, you might want to check with other groups to see that there is no conflict
... hearing no objections lets go with Thursday/Friday

sandro: what about annotations?

tantek: do we want to conflict or specificially not conflict

<Loqi> I added a countdown for 3/16 11:45am (#5814)

<Loqi> Loqi has 394 karma

RESOLUTION: reserve Thursday/Friday for meeting at TPAC in Lisbon

eprodrom: if we are to have another meeting after that it would be close to the end of the year, do we want to schedule that?

tantek: last time we discussed we suggested SF, i'd be happy to host

eprodrom: would there be any benefit to being here?

sandro: not really other than travel rotation

AnnBass: it would be more fair since the summer one is on the west coast

tsyesika: its easier for us to get here

<Loqi> Abasset made 1 edit to Socialwg/2016-03-16

<Loqi> Abasset made 1 edit to Socialwg/2016-06-07

<Loqi> Wilkie made 1 edit to Socialwg/2016-06-07

eprodrom: i'm not sure there is any benefit to meeting that close to the end of our charter

tantek: hopefully its to celebrate our completions

sandro: probably no decisions to make at that point

aaronpk: if we have to decide that before december, is it better to have it earlier?

tantek: now is the time to start planning date ranges certainly
... normally the TPAC is in early november, so its a little strange this year

eprodrom: could we sketch in a proposed F2F @MIT for early november 2016?

wilkie: we can always revisit this in june

tantek: this would give people plenty of time react

sandro: the week of Nov 14th doesn't conflict with anything (elections, holidays, etc)

tantek: lets just say early Nov so people have time to react to that

AnnBass: are we officially saying it here at MIT?

tantek: there is a proposal for that, I'm still happy to host in san fransisco too

sandro: november is far enough out we can ACTUALLY get our choice of rooms

tantek: we'll follow up on future F2F discussions with that schedule
... any other input on F2F or all other admin items?

activity streams conformance discussions


eprodrom: just for context, one of the decisions we made was to add a conformance section to AS2 before getting to CR


eprodrom: specificially chose to mimic the html conformacne section
... this task has been on my todo list for some time, so i'm glad this is done
... can we step through this? its not as long as HTML5 conformance section as it had a lot of previous versions and its not as significant here for some of those sections
... non-interactive presentation for example is too detailed for AS2
... followed structure of html prolog
... the RFC 2119 is duplicated here, we can probably remove it here or at the beginning
... in terms of the conformance classes, its a description of the various roles

<Loqi> Abasset made 2 edits to Socialwg/2016-03-16

eprodrom: at times we say implementers must vs documents must, so we call that out here
... some specific MUST and SHOULDS are called out here
... if anyone wants to queue up, please do so
... prefer this vocab over other vocabularies
... there are JSON-LD features that should not be used
... those are detailed throughout the doc

sandro: why are these all SHOULDs not MUSTs?

tantek: maybe we should make these MUSTs

eprodrom: the specific sections are SHOULDs

tantek: we should update the SHOULDS to MUSTS as this conformance section seems to say it optional

sandro: it seems rather odd to just repeat all of these from the document

eprodrom: i was trying to bring up key points for conformance

<tantek> specifically, we could updated this: "Conforming publishers should make conforming documents available according to the serialization requirements of section 2." to "Conforming publishers must make conforming documents available according to the serialization requirements of section 2."

eprodrom: there are ways within JSON-LD to specify multiple languages for example and an AS2 version of that and we are saying to not use those things specifically disallowed by the spec
... let me ask a higher level question, i think the intention was to say "what is a good AS2 document?" and i wanted to collect that all in to one section
... is that a worthwhile goal for this section?
... is that the wrong kind of effort to put in to this conformance section

tantek: i think from a spec writing side, its better to prefer a MUST instead of a SHOULD

eprodrom: using 1.0 syntax for some of the properties is a MAY (from memory)

sandro: if you are as2 with some compatibility for as1 processors, are you "conformant"

tantek: is there a transition section in the spec?

eprodrom: yes
... the idea is there are 2 media types

sandro: sounds like as2 media MUST NOT have as1 content

tantek: if you can't come up with a specific reason for a SHOULD, it must be a MUST

rhiaro: if alternatives exist AS2 terms must be present, but you can have non as2 if there is no equivalent or as long as the equivalent is there

(some discussion about exact phrasing)


(i cannot scribe live editing sanely)


(edits to make last section more clearly)

sandro: have other editors been thinking about conformance clauses

aaronpk: webmention and micropub have short conformance clauses but nothing this complex

tantek: can acitivypub use that

<aaronpk> micropub conformance section:

sandro: its much more complex for document formats

<tantek> consider s/Conforming publishers should make conforming documents/Conforming publishers must make conforming documents

eprodrom: done editing, it now says use the equivalents but also uses parallels

tantek: i had one more suggestion, in the publishers section

sandro: we haven't gotten there

eprodrom: there may be some other examples of documents we can add

<Loqi> Aaronpk made 1 edit to Socialwg/2016-03-16

eprodrom: i give a description of implementations as that word is used a lot

tantek: we can remove "human processes"

<tantek> * human processes

sandro: maybe specify they two types of implementations are publishers and consumers
... i think we can say MUST consider, its not really any way to test or enforce that
... at least make it stronger

tantek: human impact of the word MUST at least

eprodrom: since most of the publishing requirements are in the spec there is not much need for a large section here other than to say they have to create a document that conforms
... the final section is on consumers
... there are things that are MUSTs for them such as continuing to process if they hit things they don't understand

sandro: have you had feedback from jasnell_ on this?

eprodrom: no

tantek: has anyone heard from james?

<jasnell_> I'm here. somewhat ;-)

eprodrom: lets make sure we get his sign off.

<jasnell_> haven't been able to follow along with the conversation for a bit

<eprodrom> Oh hey!

<eprodrom> jasnell_, we've just discussed waiting for your review on

tantek: do you think there is enough experience with how feed-readers treat other formats, to provide a base of how readers MUST treat AS2?

eprodrom: thats true we do have some experience from nearby domains we could borrow from

tantek: we could at least say "if you are building a feed reader here is your list of MUSTs"
... it would be a specific subsection for feed readers

Karli: do we mean an RSS reader here or more like a social client

<jasnell_> the conformance page looks ok, pretty light in terms of actual conformance rules but that's to be expected

tantek: its better to start more like an RSS reader, and we can expand later
... i'm not sure we have the experience to do that for social networks yet

<jasnell_> for producers, something should be said about conforming to the rules of each property. For instance, "updated" and "published" must be iso8601 date-time

eprodrom: it would take me a while to do examples for all of these classes

tantek: i'm talking about only for 1 specifically, not for any others

<jasnell_> "name" and "nameMap" must not contain markup, and consumers must not treat it as markup

<jasnell_> "summary"/"content"/"summaryMap"/"contentMap", however, are HTML and consumers should treat them as such, etc

eprodrom: its a very good example to use too
... i'm happy to do that, maybe even for tomorrow

Action eprodrom to add a section on feed readers to AS2 conformance section

<trackbot> Created ACTION-86 - Add a section on feed readers to as2 conformance section [on Evan Prodromou - due 2016-03-23].

tantek: i agree with jasnell_'s comments


<jasnell_> there are other examples, I'll see if I can come up with a more exhaustive list

<tantek> and adding that consumers MUST treat them as such

eprodrom: if you should through this document for capital MUST and Capital SHOULD, we have 30 must 31 should, 18 for may
... would recaping those be good for this section?

tantek: no, those requirements should be hardened

<jasnell_> but they include things like: on as:Link objects, the "id" property must not be considered to be the actual URL to the object, the "href" property must be used for that. The "id" and "href" can have the same value, but they serve different purposes

<eprodrom> jasnell_: I count about 30-40 each of MUST, SHOULD, MAY in the text

<eprodrom> Is the conformance section a good place to recapitulate them?

<jasnell_> not everything is captured as MUST/MAY

<eprodrom> Or should be refer to them by reference ("serialization as in section 2")

<eprodrom> jasnell_, such as for example, range of properties on different classes

<jasnell_> I'll take some time later on today and tomorrow to write up as many of these types of rules as I can

<eprodrom> jasnell_: feel free to edit the wiki page

tantek: if those are specific properties, edit the spec
... those requirements should go with the properties

<jasnell_> btw, my apologies for being awol the past couple of months, the Node community stuff has been keeping me quite occupied

tantek: later in the spec you can say "consumers MUST" and we know what a consumer is

<eprodrom> jasnell_: no problem, just glad to get it here

<eprodrom> jasnell_: let's powwow one-on-one for further editorial points

<jasnell_> +1

<jasnell_> if you would, send me a calendar invite for a time that works best for you

eprodrom: are there other things you would expect here or that are confusing?

sandro: i think this is good, the sort of lawyer perspective is what is someone going to get away with because of something we missed here? I think this is the right track but we can't guess what all those loopholes are in the future.

tantek: this is also going to feed into the test suite, any must can be captured in a test

eprodrom: is there a decision point we can make like "include a conformance section" i think we've already approved that, can we add this now as a first draft

sandro: i think we can do that and hope anyone spots any fixes before we publish again

tantek: i think this is good to go in now and i'd like to see it published sooner than later but we can talk about that in publishing schedule later
... shall we take a break?

rhiaro: lunch is here, half an hour early

tantek: lets do issue labels section then break for lunch

Issue Labels

sandro: issues labels on github is an over-contraint problem, you'd like it to be overly clear but issues have to serve several different reasons
... sometimes they are used for w3c process, sometimes for editors, sometimes for chairs


sandro: i tried to put that all togeter in to a repo of mine
... commentor stuff is probably w3c process
... mostly these are on closed issues
... the Waiting ones are the ones the chairs would look at
... these are things that would show up on the agenda

the greyed out ones are ones can sort of ignore

eprodrom: i have two questions on this, we have a large number of repos on this

sandro: i have a script that can put all of these on a repo

eprodrom: this is for a spec repo, what about software / test suite

sandro: i assumed a spec

eprodrom: is there any reason not to start doing this immediately?

tantek: i'd like to see fewer

dmitriz: "editorial" vs "editorial: spec is ambiguous" is there a difference there?

sandro: editorial is not important change, spec is ambiguous is a little more dangerous

tantek: that sounds not editorial

Karli: with owncloud we start with fewer and add them as we need them, its also not clear here who sets these labels

sandro: agreed though i'm not sure we have a good answer to that
... someone with github permissions

eprodrom: my understanding is that some of these are anticipating questions of CR

sandro: for example we have 176 closed comments on AS2, we would want to have them all organized

tantek: i'm pretty sure thats for last calls etc, so we don't have to go through all old comments

sandro: i don't think thats true

tantek: in other groups after we went to CR we added them to "disposition of comments" but not before we went to CR
... it was "have you actually responded to all comments after the last call for comments"

sandro: okay, maybe we don't have to do that, We SHOULD do that..

tantek: i don't think we should, its a waste of our time

dmitriz: in the interest of reducing number of tags, 'waiting for' could be grouped

aaronpk: as an editor i like to mark things specifically for the group

tantek: i think we should limit it to those that help the W3C Process, and add it as needed

sandro: for example james at last f2f, he had to organize them for us
... this would give us that organization

cwebber2: it would help us get some idea of priority setting

<cwebber2> :)

eprodrom: i don't think we are going to come up with an agreed upon list before lunch and i want to eat

sandro: other idea is to add these to the repos and remove any unused ones after a month

shevski: we could vote like on the voice
... i think that list is too long, i'd be interested to know which are the most important
... I wouldn't know what are the most important ones

eprodrom: can i suggest moving this to a wiki page and edit it down there?

aaronpk: i would need a description of what each of these are and what they are all needed for
... maybe we just start with a list of these as the official list and if i need to add it, i look up the list and add it as needed to github

sandro: i can give you the CURL to add those with those colors

aaronpk: that does not sound like a lot of work for me
... without these labels the editor has a lot more work
... these labels would help other people to work on the spec

eprodrom: when do we move our repos under the w3c space on github?

tantek: its not a requirement

sandro: i'll add something of a description of them all to the wiki with their colors

tantek: i'm happy to add what i think is important of these

AnnBass: i printed out and ran through editorial changes for the 3 specs on the reading list

i'll give those to editors

Action sandro to add labels to the wiki with short description of each

<trackbot> Created ACTION-87 - Add labels to the wiki with short description of each [on Sandro Hawke - due 2016-03-23].

<annbass_> Jasnell: I have a paper print of AS, with a few "English" edits ... Would it work for me to give it to Arnaud next week, to give to you?

<annbass_> (Paper is easier for me to edit in this manner, than online ... For making suggestions to you)

<wilkie> reconvene at 1pm

tantek: we are on lunch until 1pm to do as2 validator and taking as2 to CR both of which are led by evan so i'll chair and then we can swap off for the rest of the day

<Loqi> I added a countdown for 3/16 1:00pm (#5815)

<jasnell_> hey, sorry, got pulled away for a bit

<jasnell_> annbass_ : I'm not likely to see arnaud soon, but if you'd like to scan and email me a copy I'll make the edits

<annbass_> Seems like a lottaa

<annbass_> Whoops ... A lotta pages to scan ... Maybe give me your snail mail address in private, and I'll do that?

<annbass_> (23 pages)

<Loqi> Tantekelik made 2 edits to Socialwg/2016-03-16

<Loqi> Sandro made 1 edit to Socialwg/2016-03-16

<aaronpk> Arnaud: can you hear now?

<Arnaud> nope

<Arnaud> in the meantime I tried another computer and had the same results so I'm pretty sure the problem isn't on my side

<aaronpk> just refreshed. now?

<Arnaud> yes, it works!

<Arnaud> thanks

<Arnaud> hi :)

<tantek> eprodrom is getting setup to demo the AS2 validator

<tantek> we need a scribe for the afternoon

<tantek> thanks cwebber2 for scribing

<tantek> scribe: cwebber2

<ben_thatmustbeme> scribenick: cwebber2

<Arnaud> thanks aaronpk!

<Arnaud> the sound is actually pretty good

<aaronpk> great

Demo: Activity Streams 2.0 Validator (Evan Prodromou)

<aaronpk> you can hear evan okay? he's on the opposite side of the room as the mic

<Arnaud> very well

eprodrom: well I'll get started then. The point of this demonstration from a pull-back, one of the things we have to do as we bring AS2 to recommendation is have a test suite. But one question is what that means for a document format specification. We've taken it to mean two parts
... 1 a set of document formats that we expect consumers to consumes, that's in the activitystreams-test-documents, composed of examples from AS Core and Vocabulary specifications, as well as jasnell's javascript stuff for AS2
... it's about 100 documents, maybe 200
... a number of different documents
... our expectation is those writing consumer implementations should be able to consume this, and if their thing explodes they know they have a problem
... so it's a fairly low impact test suite for consumers

sandro: it's 200

eprodrom: 200 docs, thanks. on public side we need to validate ? so we decided to build a validator
... so this is a way to test your documents. the as2 validator is written in node.js, canonical version at, if there's a reason to run another version we can do that too, software is under w3c software license


sandro: it's not linked from

eprodrom: I'll drop it on irc (^^-- above)
... 2 ways to submit to validator: paste the URL into the URL area; if you see there's things at GitHub let's get the raw version of documents
... there are bad parts, I think it's served as plaintext, but let's just say we've got this
... when we hit the url we see this validation report, we see that it's from one of the example documents
... I'm not crazy about this reports, there are 3 errors which I think is too high, but I think it's a way to do that submission
... the problem is that we've got several objects in here without a required name property

tantek: is name or name map a must?

eprodrom: yes it's a MUST
... it's one thing that comes up, it's a MUST with few examples in the spec
... the other thing that we could do here is to copy paste it in here, take this same document, copy pasta, run another validation
... and it'll run validation again
... you can also upload a file, so you can say hey look, I'm in the test documents, how useful, upload that, validate it
... and there we go. lastly, you can use the validate endpoint as an api endpoint and via your fave programming language or curl on the command line, you can do that
... let me bring up a terminal...
... so this is just a curl command line http client, will shoot it off to the endpoint
... the endpoint returns first a list of notes, similar to what we see in the html interface, also returns the input that it received so you can verify you actually got what you thought you did
... so that is a way to do some validation from the command line... I use this to validate all the documents that are in the test document suite
... so that's a relatively easy shell

sandro: when I try that I get html with a link to the validator

eprodrom: yeah do it with https
... you have to do something to tell curl to do the redirects if you use http
... I think this should follow a lot of the use cases, I think there may be other ones, but api endpoint is kind of a failsafe on the rest
... there are 4 ways to submit a document
... speaking of validation, one of the things that's interesting about AS2 is that it's very permissive. an empty javascript object is a valid AS2 document
... no properties are absolutely property, only one is that you MUST have a name on certain kinds of documents
... despite that, there are some better and worse AS2 docs
... so we might tell users there are things we'd like to see
... I did this by showing a heirarchy of notes
... I'm not sure where this comes from but I know it from syslog style errors... ERROR, WARNING, NOTICE, INFO
... MUST should do error, should should do warning, style issues should be notices or informational, maybe a style thing
... notice is something you should probably change, and info is just info
... this seems to be working pretty well... I've implemented almost all the MUST, most of the SHOULDs, and a few of the MAYs and optional properties I haven't yet followed up on

for most of the properties and thet ypes we have in AS2.0 the domain and range of propertis are if you have an actor for the activity what are the things it could be

scribe: so domain and range of properties, required properties, recommended properties, things allowed and not

sandro: so test documents are those all supposed to be OK documents? or should they be categorized with some as errors, warnings

eprodrom: good question, all of them are "these should work"

sandro: except maybe that one that wasn't ;)

eprodrom: yes maybe that doc or our spec that needs fixing

sandro: it might be nice to manually sort so we can test this

tantek: it's realtively stable at this point, could we add links for all warnings and errors?

eprodrom: good idea

tantek: that way if someone gets an error they can click it to see how to fix it

eprodrom: yeah
... there are two views, you can see submitting or validation report, at the moment it's pretty plain janes but

tantek: can you show url based result again

eprodrom: yes, you can also tell one thing it's not doing here is complaining that it's being served as text/plain, which is one more thing it can do

tantek: that's great, here's a thing you can share

eprodrom: right right
... you can do url=${document location}

tantek: do you have any live sites on the web you can link to

eprodrom: not yet

tantek: valid challenge to working group: who can get a valid stream on their site first

sandro: are there any as1 -> as2 converters

<tsyesika> AS2 document on the web :)

eprodrom: any other questoins?

tantek: mostly that it's awesome
... one nice thing is our discussions are not subject to rfc2019

eprodrom: I'd love help, styling would be nice, I'm not sure, maybe tabs

<Arnaud> I agree, nice progress!

<rhiaro> My photo albums:

tantek: next topic is taking AS2 to CR, which is also yours eprodrom :)

Taking AS2 to CR (Evan)

eprodrom: what I'd like to do is my understanding is that we had three things we needed to do to get AS2 to CR state
... these three things were to first resolve (blocking) issues, second provide a test suite, third provide a conformance section
... I think as it stands we have first drafts or early versions of those three things
... I may need to confirm that about the issues but at least in terms of test suite we have what we've kind of laid out as in terms of test suite, and we have conformance section in a draft state

tantek: for test suite do you have at least one test per feature of the spec

eprodrom: would that be like for each type in the vocab have on document for each one
... so for properties we have at least one of each
... a property in as2 can't be an empty array for example
... here's a document with an empty array, throw an error

tantek: do you have a test for an assertion for each part of the spec

eprodrom: since we have tests from documentation examples, we have pretty good coverage

tantek: that's ok, having a comprehensive test suite is not a requirement to enter CR (but it is to exit)

sandro: I don't remember what the granularity was

tantek: was it raised as an issue to define granularity

sandro: yes

eprodrom: we went down an interesting path with that one, and it got caught up with the conformance classes

tantek: we have conformance section, have test suite with a lot of cases covered, so that leaves the open issues
... so where are we in open issues

sandro: we have 12, and I won't snark that they aren't labeled ;)
... one of them is editorial but it's not my version of editorial!

eprodrom: we have 12, one is "why is as2 so bloated"

sandro: it has more in the issue than the title sounds like

eprodrom: right
... so we have quite a few here, some are questions, some may be... I'm not sure if we should mark these as blocking or not

<Arnaud> issues like "why is as2 so bloated?" should be closed with a comment stating that this is not actionable, if any particular part is questioned they should be listed in a separate issue

tantek: the question is for you, how much time do you think you need to resolve these, and which of these do you want to ask for some group discussion on to move them forward

<aaronpk> "this is not actionable" is my favorite way to close issues :P

eprodrom: not ready to do that today, I might be able to tomorrow, I'd like to run by james, would be great if we could run from 12 to 6

tantek: or 0

eprodrom: right or 0

tantek: I'd like to get as close as we can to that by end of this meeting
... so given what you said about not willing to categorize, can you be by tomorrow morning?
... we have agenda item scheduling and... we could do that first thing
... or you might come back tomorrow and say you resolved all

eprodrom: ha ha ha

sandro: james may participate remotely, maybe we can get an answer from him about ??

tantek: determine which you need group help for, then can james be present or is he happy to delegate to you about the issues

eprodrom: fine, good

tantek: otherwise will add to agenda for tomorrow morning; go through AS2 issues

eprodrom: with intent to close!

tantek: will block out an hour and a half for that

eprodrom: could I ask for rest of group to look through this? may speed discussion if people understand what topics are

tantek: consider this explicit call for commentary
... some of these like conformance clause you reasonably fixed

eprodrom: yeah we were almost done by last f2f, but then we had f2f explosion with new issues
... this one is interesting can try to take that on today

tantek: every document I've seen has CR exit criteria in spec or include it inline

shevski: why not have it be in the official w3c-social?

aaronpk: it isn't a requirement to do so, some required moving to their own group

<tantek> ack

annbass: question is is there some process where it should go ahead of CR, I could easily see one person's repo goes south or something

tantek: or that's one possible org issue

aaronpk: before github was used more mailing list was only place to do that, but now github is using more
... so I wonder if we should be archiving that discussion

annbass: that might be an advisory board thing

<tantek> There is the W3C-wide document on use of GitHub:

tantek: repo issue is possibly helpful

aaronpk: that's possibly useful around patent stuff

tantek: lots of good reasons to do that

<aaronpk> it seems there is not a good answer to the particular question of archiving the github issues discussions

<tantek> break for 10 minutes

<tantek> break until 14:10 EDT

<tantek> Loqi ^^^

<rhiaro> eprodrom: does the validator send an accept header?

<wilkie> come back at 14:10 EDT

<Loqi> I added a countdown for 3/16 11:10am (#5817)

<rhiaro> I could look at the source but it's easier to ask :)

<Loqi> Tantekelik made 2 edits to Socialwg/2016-03-16

<Loqi> come back

<Loqi> Countdown set by wilkie on 3/16/16 at 10:59am

<aaronpk> you're the only one on the talky right now anyway :P

<rhiaro> scribe: rhiaro

<ben_thatmustbeme> scribenick: rhiaro


<tantek> chair: eprodrom

Demo of - mediagoblin federation

tsyesika: this is a predecessor to activitypub, the pumpio as1 stuff
... Going to add a photo onto media goblin and send it to a pump io user
... *does so*

<bengo> so good!

tsyesika: it has federated from mediagoblin to pumpio
... Can post a reply in pumpio
... Back into mediagoblin, the reply should show
... *it does*

Everyone: applause

tsyesika: Coming up in the release after next
... Any questions?

eprodrom: Bravo! Couple of questions..
... How does it work for idfferent media types?
... Media goblin is generous with media types

tyesika: Media goblin has a todo to support video and audio in the api for federation. Currently only images federated
... It will support them in the same way as images. Serialise the video or audio as how as1 defines them, and do the same post request to the inbox

eprodrom: About likes... do they translate to media goblin?
... If you go to the pumpio page and click like on it

tsyesika: they will work by the time I finish mediagoblin
... Things are still a bit broken

eprodrom: What do you think next steps are with this?
... with federation of mediagoblin?

tsyesika: Currently we don't have comments that you make in mediagoblin federated *back*. All activities need to federate and there will be support of like and the others.
... And then probably the media types as you brought up
... One of the reasons media goblin is good is support of media types
... And eventually having a up to date implementation of activitypub

<Loqi> Tantekelik made 1 edit to Socialwg/2016-03-16

cwebber2: I think it's interesting for the group to know that tsyesika has been working on this for a long time, and you might wonder why it took so long to get to this point, and I think it would be interesting for you to explain what the challenges are

tsyesika: The overwhelming biggest thing for mediagoblin why it took so long is because of the database structure
... Mediagoblin uses sql, postgres or sqlite, and it was designed prior to federation and the pump api
... having things like comments that can be on other things, comments on comments and other things being in collections, involves creating generic keys
... something something referential integrity
... Lots of challenges implementing federation on an existing implementation

eprodrom: That's a really interesting point, pumpio uses document databases, so couchbase or mongo or redis, but it's relatively easy to wedge little bits of data into places that they weren't intended
... which is harder to do with an sql database

cwebber2: we're theoretically advocating ... and this applies to others.. it's a general problem when you have assumptinos about what types of things are responses to other types of things
... Like comments ony being on media, that was an easy assumption, but the rest of the world might not assume that
... We don't have an answer for constraints in federation
... and this will be a challenge for owncloud as well. We haven't talked about this in the gorup
... How to convert existing applications that were designed before you planned to federate, and move them into the federation wrold
... We're managing it with media goblin thanks to jessica, but not everyone will have the resources for that
... WHat can we do to help people know and prepare fo rdoing this?
... Do we permit organisations to set these kinds of constraints?

Karli: database structure is part of that, but not every social network works the same
... Facebook and g+ and twitter all have apis that work differently
... twitter has obvious limitations with the size of the text, if you can like, if there are comments or retweets which don't exist in other social networks
... It's a huge question. Good to have an api that you can use to model all kinds of social interactions, but does it mena that every social network has to support everything that every other social network supports?
... You could drop things that are not supported

eprodrom: that's probably a great way to handle it
... THere might be other ways to do it too
... You oculd have a little grey image icon 'there is something here I don't understand'
... like a puzzle piece in html documents
... some embedded stuff that doesn't make sense
... pumpio just throws stuff out

tantek: the experience with webmention is that there are situations where having defined fallback behaviour is useful and implementable. And also sometimes just dropping stuff works. Very case by case.
... By starting with.. just get comments working. Suddenly everything else can have a fallback that can be interpreted as a comment
... So when people started doing lieks via webmention, by including a summary in your like that says 'so and so likes this'
... if you understand likes you ignore the summary and just increment your counter or whatever
... but if you don't, you can display it as a comment
... so there's no new implementation needed

Karli: so it's the responsibilty of the consumer to ignore or translate something?

tantek: we didn't have to, because comments were designed in such a way to handle summary/content/etc, that later interactions could provide fallback content like a summary, so that something that only implements comments doesn't have to do anything new

aaronpk: reacji is a great example
... Someone posts a single emoji reply
... Slack formalised the name
... You choose an emoji character as a reply
... you see on github a lot
... You end up with long comment threads full of emoji
... Slack did a thing where they showed the emoji with the number of people who reacted with that
... So we're starting to experiment with that, so ifyou don't do anything an dsomeone posts an emoji reply you show it as a comment, but if you want to you can not show it as a comment you can pull it up and create a counter
... the like example seems trivial becasue everyoen knows what it is, but now it's happening again

cwebber2: one of the things that's interesting about ... you must have started with a database structure that still allows that
... but existing implementatkons like mediagoblin start with databases where it's still difficult to even do fallbacks in that way
... if your table only contains these fields, what do you do as a fallback? If you have hard coded links between tables, you need to be able to make that generic enough to do fallbacks
... In indieweb you have this view that everything is a post and you can adapt that. But other things don't have that assumption in their worldview
... So how do we build on these different world views?
... You would be surprised at the different world views, even trying to be accommodating for fallbacks

tantek: the more you can minimise those assumptions the better

cwebber2: that's not necesarily true for an organisation that already came in iwth its own assumptions

tantek: from a spec point of view. Of course organisations have their own assumptions. But with a spec, the more you can minimise the assumption sthat you're asking peopel to take on, the less work it is to make it compatible

shevski: the problem also is that this stuff is going to continue to evolve
... reactionmojis
... it's going to be really fluid. You don't want want to get into a race of being compatible now, because it's not going to stay around. It needs to be generic and extensible, and quite basic int erms of standards

tantek: we've done both, compatible with now and generic
... the web took off because every single piece was cmpatibile with what already existed
... you could serve html over ftp, or plain text over http

eprodrom: I'd like to bring it back to the pumpio and mediagoblin federation
... There is a lot of pumpio federation process that's kind of string and chewing gum
... THe oauth key discovery process is like from an older version of oauth2 that's no longer compatible
... There's the dial back stuff
... There's the webfinger stuff in there
... And probably other bits, hacked to make them work
... Do you think that where we're going with activitypub is going to .. what's th erelationship with all that stuff and where we're going with activitypub?

tsyesika: good question
... ap at the moment doesn't specify too much of those details
... We've decided oauth2

cwebber2: we were also told not to pick a stance there

tsyesika: so it's vague, so currently they would be incompatible because you can implement oauth2 in several ways

<bengo> <3 OIDC (which is a more specified flavor of oauth2)

tsyesika: the discovery stuff, we do have stuff but we think it will change

cwebber2: we already agreed on some change. We just haven't written yet

<bengo> including discovery, dynamic client registration, etc

tsyesika: so doesn't specify all of that, but as a downside will lead to incompatible implemenations without specifying

cwebber2: do you think there's anything other than auth?

<bengo> Fair. Communities of different size tho

tsyesika: i don't think anything other that's major

cwebber2: that's the big thing that's a known problem. If it's a constraint then it's a constraint

eprodrom: I see two ways of doing that. One is to include authenticaiton and say this is how you get an oauth 2 key in order to make these calls, and here are the steps
... Another way to maintain compatibility is to have a suite of specifications, AP specifically talks about getting your activities back and forth, and then a discovery specification and an authentication specification or something like that, which would be short

cwebber2: I think discovery would be smooth put in there, but authentication I would be enthusiastic about saying there are auxilliary specs, but they might not make it to rec within this group's lifetime, but we have something that we can point people to. I think that's a good route.

eprodrom: this is a lot of trouble to go to to end up with incompatible implementations
... As one of the other people who has done this, I know how much trouble this was
... I'm really impressed
... Any other quesitons?

tantek: how much work would it be to update these implementations to use as2/ap

tsyesika: I don't think.. it's still significant amount of work, but not too bad

cwebber2: lots of things close, major things that are different are.. moving from as1 to as2 is not very hard, but needs to be done.
... We need to change over the way it does discovery
... And ... those are the biggest things
... I think there are a lot more small things, but most things are close

tsyesika: I don't think it's too bad

cwebber2: shouldnt' be as much work as it was to restructure the database

Going to be far less work to get to that point

tantek: When do you think you'll have an implementaiton of the current version of ap?

cwebber2: this depends on resources within the project. We were lucky that we could bring jessica on with a crowdfunding campaign
... now that money is ending and jessica got a job that is supprotive of her work in this group luckily
... But I'm currently at Stripe retreat, but after this we don't know what our resources are. Might be volunteers
... Really hard to say
... THe stuff that you're seeing here is for a release in the next week
... Then we're polishing 1.0, then we'll start moving things towards ap
... That's basically the state of the wrold
... How long that will take... I'm already bad at estimations... so doing that for a project where I have no idea what our resources are, I'm not going to bother

tantek: so you think you can release with this implementation?

cwebber2: yes, with notes that should expect federation stuff to change, and in the interim it will federate with pumpio
... Users are expecting that first stage of federation will be at this level

tantek: maintain back compat?

cwebber2: I'm more worried about authentication for that
... activitystreams stuff is manageable
... Not that worried
... Don't want jessica's work to not get out there
... Really want working federation otu the door

tantek: several generations of frozen federation suites out there that are being left behind. Like diaspora, federating amongst itslef and nobody else. And GNU social. Thousands of users, but not federating with others, only themselves.
... Movement has been slow.
... This is why I'm concerned that if you're going to make a release that includes this work will it end up like one of those, or will it move quickly and break things?

cwebber2: I'm hopi;ng move quickly and break things
... our goal is to support the work in this group
... We've been clear about that with users

<Loqi> can confirm that lunch will be delivered

<Loqi> Countdown set by rhiaro on 3/16/16 at 7:37am

cwebber2: And some people in disapora are paying attention to what we're doing here

eprodrom: Can I challenge us to do a similar demo to this with AP in portland?

cwebber2: We can certainly try
... But it's going to be a tumultuous time in our lives
... We will try, but no promises. Set your expecations accordingly.

Karli: I'm still part outsider.. but at owncloud we want to implement something soon. But we need something that is relatively stable
... That we can see a path.

tantek: what level of breaking new ground vs. interoperating with a few things vs interoperating with a lot ofthings are you comfortable with?

Karli: We already implemented federation for sharing files 1.5 years ago, and several of the steps here
... Version 3 of our api which is backward compatible, and 1.5 years later it's something 'stable' or 1.0
... So no problem with adapting to changes
... But as I understand this group, there are some really different approaches, like activitypub and solid stuff and other approaches, which are completely different
... We can't implement 2 or 3 different approaches

eprodrom: That's a good segue

Demo of micropub CRUD

<Arnaud> it works!

aaronpk: Since the last time we talked I've worked on spec and implementation to support editing and deleting posts
... So I can show the whole workflow
... Quill is a micropub client that I wrote that currently only supports creating
... My website is the micropub server which uspports full CRUD
... And there's a micropub client built into my site
... *creates post in Quill*

<KevinMarks> is empty - are you doing demos somewhere else I can see?

aaronpk: *shows Quill output*

      • please hold for technical issues***
      • elevator music ***

<tantek> aaronpk fiddles with talky computer

      • and we're back! ***

<KevinMarks> I can hear, but screen is fuzzy

<KevinMarks> screenshare instead of cam?

aaronpk: So, created a post. It's on my site here

<KevinMarks> I suspect screensharing needs even more RAM than cam

aaronpk: Location sent with browser location api to micropub request

      • everything crashes ***
      • uncrash ***

aaronpk: When I'm logged into my site, I have a menu bar which lets me edit
... Loads a javascript micropub client
... Because I'm logged in my site generates an access token the client can pull out of the page
... I gave myself a couple of fields I commonly want to edit: tags, syndication urls, date
... So I can add tags, hit save
... What's happening is the js app is talking to my micropub endpoint directly, no other serverside componant at all
... *does so*
... In order to populate this interface with existing tags etc, it uses a new part of micropub which is the R in CRUD which lets the client request specific properties
... I don't need to load the entire content of the post to edit
... So it reuqests just the pieces that it needs
... There's no way to edit the text of the post with this interface, don't have a client for thsi yet
... Turns out that I don't often edit the text. Wanted to optimise this for the most common edits
... Do plan on building a full editor for html posts in Quill
... Also have a delete button

<KevinMarks> why is videoconferencing still so flaky? I had this working in 1999 with tiny amounts of RAM and it ran for weeks at a time

<cwebber2> KevinMarks: let's move back to webcams with pages that refresh every 30 seconds on a jpeg link

aaronpk: JS client could drop in with one file. Only special thing is how to get the access token, but I don't see why other people can't reuse it
... I could open source it
... There's not a lot of code
... Here's the micropub request in jquery
... The rest is just ui stuff

<KevinMarks> I'd accept multicast RTP

<KevinMarks> (audio has gone)

<annbass> sigh .. trying to get it going

<annbass> is sound back?

<eprodrom_> jasnell: so, big favour to ask you

<bengo> sound just came on for me

<bengo> talky is good

Demo of events and webmention

<annbass> ok .. arnaud and kevinmarks ... you can hear?

<Arnaud> yes, thanks

aaronpk: Here's a new interface in quill for posting events. Needs cleaning up, but minimal and works
... *posts event*

<KevinMarks> and the gpu just coughed a hirball and blacked out all my browser windows

<KevinMarks> yay, i can hear again

aaronpk: Event posted to my site, and appears in Woodwind, a reader
... The reader recognises that this is an event post and adds these additional rsvp buttons

<annbass> s/clenaing/cleaning/

aaronpk: I authenticated with the reader, and during that process I granted it an access token that lets it become a micropub client to my site
... So when I click a star on a post, it makes a micropub request to my site
... Similarly, when I click an rsvp button it makes an rsvp post on my site
... Also a micropub request
... So I see the rsvp in the events feed on my site
... created in quill, quill added it to my site, woodwind read it from my site, woodwind created a new post to rsvp and posted that to my site

shevski: in what ways can I rsvp to your event?

aaronpk: You can rsvp by writing a post on your site htat's an rsvp post and send a webmention to this page
... and you'll show up in that list

shevski: any other way? Only way is for me to create a post on my site using...?

aaronpk: doesn't matter how you create as long as it iends with with html + microformats

shevski: Why can't I go on your site and click a button?

aaronpk: you want me to host your rsvps?

shevski: or have that as an option
... it's a little hardcore

aaronpk: my site is not a general purpose events site, it's just my site. I also don't have a comments form.

annbass: *trying to send rsvp from known*

dmitriz: is there any access control? Can events be private?

aaronpk: not right now, planning to

dmitriz: haven't had chance to look at the spec, how does the recipient of the webmention discover that is of type event

aaronpk: here's the parsed microformats of the rsvp post. has in-reply-to and rsvp properties
... that's enough that my site recognises that it's an rsvp
... THe event itself looks like this, which has type: h-event, and start and end date
... The post type discovery spec tells you how to get from amicroformats object to something else

dmitriz: you don't have access control on this site, but is there provision for private and public?

aaronpk: what you're talking about would be more on the reading and consuming parts of the spec
... THe way that woodwind found the post, you would need to have a way to authenticate to begin with
... We don't have anything.. woodwind found this via PuSH and polling the feed. There isn't a mechanism in PuSH to support private posts. But it's on the consuming side rather than the micropub side
... Similarly in webmention we need a way to verify things that are private

Demo of rsvp with Falcon

tantek: I have an rsvp post that I"m going to be sending to aaron's event
... **php warnings, blames aaronpk**
... Is showing debugging info about finding webmention endpoints
... posts, reloads aaron's event post
... rsvp is there
... and on my site
... and on twitter
... using twitter's proprietary api

eprodrom: 20 minute break to recharge
... (recharge parking meter / people)

<wilkie> reconvene at 3:40pm EDT

<Loqi> I added a countdown for 3/16 12:40pm (#5818)

<aaronpk> Arnaud: haha i never considered that!

<aaronpk> is it not used as a verb in french?

<jasnell> eprodrom_: favor?

<KevinMarks> well the R part is the verb Respondez S'il Vous Plait - "respond if it pleases you"

<KevinMarks> but RSVP becomes a noun, and then a verb becasue that's how English imports external code

<Loqi> reconvene

<Loqi> Countdown set by wilkie on 3/16/16 at 12:18pm

<eprodrom_> jasnell: so, the only thing keeping us from voting to take AS2 to CR is that there are a number of open issues on the repo

<jasnell> ok, I weighed in on most of them

<jasnell> just need to know how the WG chooses to resolve those

<eprodrom_> Perfect

<eprodrom_> So, the favor was, "Please weigh in on them by tomorrow morning so we have an idea where you stand."

<eprodrom_> It's likely that unless there's some intense debate tomorrow we'll default to resolving per your recommendation

<Arnaud> re: RSVP, KevinMarks got it right :)

<aaronpk> is "RSVP" as an abbreviation used in french?

<Arnaud> it's also that it seems to be used both for the question and the response in english, that's the most confusing part to me

<Arnaud> yes, it is used as an abbreviation but only at the bottom of an invitation or something like that

<aaronpk> ah yes. in english I would send you an RSVP


<Arnaud> in French RSVP is merely used as the prompt/question

demo of rsvps

<Arnaud> then you would just talk about "responses"

ben_thatmust: *shows rsvp demo, same as aaronpk's*
... Now a reactji - emoji response
... post with a single emoji in it is the original post
... but the response shows as a count
... *it is a poop*
... A bunch of others from testing

eprodrom: it's not a different post type, it's just ap ost that has a single emoji? So you're looking in the content?

ben_thatmust: yes, I'm parsing out if it's a single emoji. Much harder than I expected
... An emoji is a bunch of characters, or two letters that form a flag, which is a single emoji
... Any questions?
... *annbass attempts to send poopji*

<tantek> annbass, can you post a reply?

<Loqi> reconvene

<Loqi> Countdown set by wilkie on 3/16/16 at 9:03am

Owncloud demo

<eprodrom_> "*it is a poop*"

<wilkie> Loqi: what is that

<Loqi> Eprodrom made 1 edit to Socialwg/2016-03-16

Karli: owncloud is not technically a social network
... First, install owncloud server - unzipping a file
... update some permissions
... and voila
... lots in to
... can deploy anywhere that supports php
... desktop and mobile clients are synchronised for files and favourites, comments, tags
... Owncloud is not social in itself, its for protecting my own files

<tantek> For the logs, my federated RSVP permalink from my demo:

Karli: you can have calendar, contacts, gallery, all kinds of things
... So now we have two servers, we can do federation
... Can share calendar etc with other pepole, uses caldav
... calendar is one of the most popular owncloud use cases these days
... here's a folder of documents with an example .odt file

<aaronpk> webmention implementations were: Falcon (sending), p3k's Telegraph (sending), (receiving), (sending and receiving)

Karli: here is the sharing side bar, can share with people on the same server, or I can send out links to everyone, or I can do federation
... For federation I type in a federation ID which is a host name + a username
... so I can now do federation with a user on another server

<aaronpk> micropub implementations were: p3k (server), Quill (client), Woodwind (client), (server and client)

Karli: one server pings the other server and says here is a sharing request
... on the other server, it shows a notification which lets me accept the sharing request
... and then I have... maybe... or not...
... I have the documents folder shared between the two servers
... The actual transfer of the files is done via webdav
... but we had to come up with our own protocol for the invitation process
... so this was implemented 1.5 years ago. Nowadays solid does stuff here that could have been resused but wasn't because of timing
... The social part is then if you click on this shared folder the sidebar opens, which lets you comment
... which is visible to everyone with access to this file
... and there's an activitystream which lets you see what's happening with this folder. New files, comments, changes
... And there's an overall activity feed where I can see everything that's going on on my server

sandro: does the activity feed live on the same server as the file?

Karli: yes

sandro: so if 100 people share a file, there's one feed?

Karli: everyone with access to a file or folder also sees the activity on this file or folder

sandro: if I make a comment it's sent back to the server where the file is

Karli: this is the part that's missing. We do everything with webdav. Sharing information including comments via webdav
... We could just read comments from remote servers via webdav
... but if there's a different appraoch we would implement that

dmitriz: how do yu handle permissions and access control?

Karli: if you go on a folder, sharing, there are different ways to share
... you can password protect links, allow expiry, allow read only
... expiration date google announced yesterday, but we had for 5 years

dmitriz: HAH

Karli: Other people can delete files, or add a new file in a shared folder

dmitriz: where do you store access control information?

Karli: filesystem for storing files, metadata including access control in database
... can be sqlite or mysql or postgres
... So not a social network, we're approaching this from a different direction
... Idea was for file hosting. But of course I want to share my files with someone
... But without uploading it to a social network
... All this sharing information is also present on the desktop clients
... communicates via rest api with the server to initate the sharing request with the server

shevski: if something is shared do they get a copy? what if they both change at the same time?

Karli: on the desktop and mobile side we do bi-directional syncing
... does conflict detection and you might get a conflict file if necessary
... but the federation is not synced, it's a live connection
... We have implemented our own api for the activity feed but we want to do the next version with AS2
... Means we will implement desktop and mobile clients for consuming as2

cwebber2: we should talk about if... the challenging thing with the database that we talked about.. the upload media stuff we should talk about. We have some vague discussion fo that in AP but needs work
... I'd be interested to see the direction and work together on that

Karli: Another thing from a strategic perspective.. a lot of people see owncloud as a dropbox replacement
... But I think this will evolve into a social network
... We have 8 million users
... so kind of a trojan horse

sandro: how many companies run servers?

Karli: the biggest installation that we are involved with is for half a million for universities in German
... But we just learnted that there's an installation in India with 1.1 million users
... Scaling is actually quite easy based on web technologies


sandro: dinner at 6, any objections?


shevski: I thought the status doc would be upated and we could just review it
... but it hasn't. Maybe we need to go over what we're tryign to do with it and who is going to update what

tantek: what's it for?

shevski: The idea was to partly start thinking about what are the next steps to get to CR for various things, and what do we need for the implementationr eport
... and what can we do in the time we have to grow adoption for any of these things
... Fulfilling process requirements and also the wider problem of getting more users and implementations?

tantek: it has links to implemenations?

<hhalpin> Where's the links to existing implementations?

shevski: it doesn't at the moment

sandro: webmention has a link


eprodrom: we have 2 things. One is documeting implementations, and one next steps

sandro: this is the link to all the other documents
... trying to get peopel to focus on what happens next

eprodrom: I think us as chairs need it to see what to do to push things forward

sandro: we could just talk about this for as2 right now at least for a few minutes
... I don't think we've wrapped up next steps for as2

<tantek> we already have a list of our documents on our home page: - no need for a separate page

eprodrom: Let's take til 1630 to discuss listing implementations and maintaining the list from hereon
... and from 1630 to 17-1730 discussing next steps for documents

tantek: I want to kill the document status page because I think it's useless and busy work
... We already have drafts on our homepage
... s/drafts/links to drafts

<hhalpin> We should have a page that lists implemenations per spec

<hhalpin> Given that we have soooo many specs right now

eprodrom: the purpose of this document status was leading up to this meeting, it's value goes down now we're at the meeting. We cna just talk about it

<hhalpin> The rest of the material on the page is busywork

<hhalpin> However, the list of implementations is super-important for CR exit

sandro: I find this useful to see status of documents, I don't mind where

aaronpk: what status do you mean?

<tantek> if you're editing a draft, please update with a link to your Issues page

<hhalpin> (just to repeat what I said above)

aaronpk: WD or other major ones, what are they?

eprodrom: let's take 10 minutes per draft and takl about next steps

<hhalpin> +1

eprodrom: For editors can say what they're doing, and people with experience of rec process can say what we're shooting for next

<aaronpk> just added links to issues for webmention and micropub

sandro: the output of this conversaiton si not documented anywhere besides the minutes

eprodrom: could we put this in github?

sandro: see what they are first?

eprodrom: start with AS2

AS2 status report

eprodrom: Big next steps are test suite, conformance section, and closing recent issues
... In december we were almost all closed but now we have more
... But we are going to try to close those tomorrow morning
... Hopefully we can close them quickly, most will have editor's recommendation for next steps
... Unless there's strong debate we can probably move fast. Not that I'm discouraging debate.
... At which point it's possible we could vote to transition to CR in this meeting
... Then we would publish a new updated version based on the decisions that are made tomorrow morning
... one last editorial pass, then have something ready to go to cr
... If the group says we do x, implementors implement x, and we probably on't need another vote
... Or we could do editorial and then a round of review
... Or we say the editorial is implementing what we decided and we can go to CR
... That might be a decision to make tomorrow

tantek: How many implementations do we have of current activitystreams drafts?

cwebber2: one in activipy and half in the guile one

eprodrom: I was just making a list on the activitystreams implementation page and I count 4.5
... java, javascript, activipy, validator which okay that's cheating, and..

tantek: why is it cheating?

eprodrom: is it an implementation or an implementation tool?

tantek: implementation

eprodrom: then we're at 4.5


hhalpin: helpful to have a big page with all implementations
... with implementations that are either conformant or planning to be


<eprodrom_> rhiaro: oh, now I feel like a jerk

<cwebber2> eprodrom, added!

<Loqi> Aaronpk made 1 edit to Socialwg

eprodrom: doesn't matter, will ask later :)

<Loqi> Eprodrom made 1 edit to Socialwg/DocumentStatus

<Loqi> Sandro made 1 edit to Socialwg


sandro: As we get to having the validator on the test suite for something like as2 it would be nice to see the implementations look more like this (matrix with tests and validations), for each document does it pass it, and results

<sandro> as example

sandro: A list is nice early, but by the time we have a test suite we should have more detail about conformance
... And then what's the process for getting peopel to give us results

eprodrom: for pumpio I could maybe take links for kinds of feeds on a pumpio site and link to each of those and say if they validate
... these are the kinds of feeds that we generate

sandro: I guess actually what I want is depending on the feature granularity
... is there a feed from that implementation that uses that feature

eprodrom: we need to move our implementation pages from we implement this to implementation reports

sandro: we could borrow something from other groups or roll our own
... An action item on someone who's willing to do it?

tantek: until we have a test suite we don't need this. First step, make a test suite

sandro: needs to be soon though

eprodrom: test reports to exit cr

tantek: which implementations are testable?
... a library is not really testable
... no working group I know of has used a library to exit cr
... not an implementation for any workign group I've been involved in

sandro: don't do universal generalisations

<eprodrom_> sandro: B-)

shevski: what are the formal requirements for w3c implementation report?
... I'm interested in demonstrating genuine adoption

<tantek> /me sandro isn't that a universal generalization? ;)

shevski: I don't think something should become a standard if only 3 people care about it
... If we have 5 different things we need to start talking to peopel and lining up implementations or maybe there is already 100s of peopel and companies using it
... Things in the wild

<hhalpin> I would suggest adding an estimate on the number of users per implementation.

shevski: Stuff people can be interested in, like mooncake which thoughtworks built which displays activitystreams from varous sources, they have demos, good to list user friendly things so people can see what it looks like
... so if people are looking at as2 and deciding, they can see if they want to use it

<cwebber2> hhalpin, did mooncake use the java library of AS2 or have its own

<tantek> +1

<cwebber2> hhalpin, is that another implementation?

<hhalpin> I think they re-coded it using Clojure

eprodrom: if we look at previous implementations for as2 that's definitely an outreach process and we do have a little bit of a psychological advantage to ahving 2.0, we can ask people when they're going to upgrade

<hhalpin> So you *may* have another implementation there.

eprodrom: 'why are you using the old version'?
... good pressure with a version number
... but that is a contact process
... and that might be a next step is to be reaching out

shevski: whose responsibility is it to do that? Editors for each draft?

<hhalpin> That being said, I'm not sure if they have any users per se. There was supposed to be 3 pilots: Iceland, Finland, Spain - I think Finland is interested.

sandro: we're not there yet

shevski: but for initial implementaton report?

tantek: that's produced *during* CR

<hhalpin> Email Jaako Korhonon re Finland and Natalie from Thoughtworks via Clojure details, I've been removed by W3C/ERCIM from D-CENT for almost a year so not tracking :)

eprodrom: definitely a snowball effect of seeing implementations listed

hhalpin: think it would be useful.. we don't test how many users something has.. assume it has users, not ask how many. That being said, users are good. Even though it's not formally part of the process should be something the wg takes into account
... So eg. per library list code bases that use it
... Keep track of how many users each implementation has
... Would help for people looking in from outside to see what the state of play is

<sandro> and

sandro: Looking at two other implementation reports. I don't think we shoudl worry about w3c, but what the users want
... see ^

<hhalpin> Yes, but caniuse doesn't measure "users" :)

<hhalpin> We assume all these browsers have users.

<hhalpin> Its harder in this case

sandro: Tells users what they can use to get to use as2

<wilkie> doesn't it measure browser market share?

sandro: if it's good enough for users it's good enough for w3c

<hhalpin> Still, adding guess user numbers helps

sandro: but it's a lot of work
... that includes covering extensions in theory, that's where it really starts to pay off

<Zakim> tantek, you wanted to comment on we need to start talking to people and lining up implementations or maybe there is already 100s of people and companies using it

<Loqi> Aaronpk made 1 edit to Socialwg

sandro: how many consumers consume this extension so should I produce it

<hhalpin> So, in theory in some of the SemWeb space (GRDDL comes to mind) I've seen stuff with users numbering in less than a dozen go to Rec, but I think W3C is discouraging that kind of thing.

tantek: I agree that listing things that are accesible to users is a good thing
... Good to list user friendly things so people can see what it looks like, as shevski said
... Before we ask anyone outside this WG to implement something, I would like to see implementations by people in this wg

<hhalpin> Typically a WG converges, then we do a call for implementations. Its a bit odd to do a call for implementation without some convergence!

tantek: For each draft, at least one. I think we're close to that
... I would hope we have an implementation from the group we can point to

sandro: I would hope the editors would have some implementation

cwebber2: and also with actual code

shevski: so people can see 'that is the kind of thing I want in my app, I will use this spec'
... so get a sense of something working

eprodrom: implementation and advocacy is going to become a very big issue before we go to cr
... right now we're collecting lists, next step is to collect test results as well as outreach


eprodrom: may be something we should start in the next few weeks
... but post CR, does that sound fair?

tantek: what do we need to take things to CR, then once we're in CR what do we need to do to exit?
... A test report is not required to enter CR
... For implementations,t hat's up to the wg to decide
... how much implementation verification do we want internally before we are comfortable taking something to cr
... we have a few implementations including a validator that anyone can try. We can make some claim like the validator implements 90% of AS2
... Then we can tell people that when we enter CR, to help encourage more
... Hopefully an example for those spec to follow
... If we have implementations from editors, 90%, 50% of the spec, that is useful to include as part of going to CR

Social web protocls

<bengo> err

<bengo> I can hear sounds but not really Amy

<rhiaro> Needs catching up with current state of specs it talks about

eprodrom: what's our intention with publication?

<rhiaro> Note is fine

sandro: remaining flexible depending on contents

shevski: seems like a primer

<Loqi> Tantekelik made 1 edit to Socialwg

eprodrom: we're shooting for another version in early april and will iterate on a regular basis over the coming months
... .. Webmention

aaronpk: A few more issues that I wasn't able to close myself, would like to go through those tomorrow
... With those resolved I feel like things are stable enough that there are no major blocking issues and plenty of implementations..

tantek: how many?

aaronpk: there's a list on the document. dozens? There are two roles, sending and receiving. I believe it's dozens of each
... it is documented in the spec
... enough where I feel like things are working fine
... The next thing I want to do is work on a test suite so there are tools for peopel to test their implementations on both sides
... interactive validators basically
... a lot already exists as small utilities

sandro: the small utitlities haven' treally poked the edge cases right?
... like where the endpoint is discovered?a

aaronpk: some test for xss for example, so they'll send you a webmention with a script hyou should be filtering out. kind of adjacent to the spec itself as the spec doesn't say you have to show a comment on a post

tantek: do you have a privacy and security section?

aaronpk: minimal, could be expanded

tantek: could expand to answer the privacy and security questionnaire from the TAG
... currently optional part of specs
... AB and TAG are gathering experience with it, but if you're not sure what to say about security and privacy this is one way of expanding


<ben_thatmustbeme> i'm guessing its that aaronpk

dmitriz: From what I remember from AS2 and AP specs the security and privacy ocnsiderations sections used language with 'must provide x level of privacy' and so on: how is that testable?
... how can we enforce that?

tantek: how that plays into conformance requirements is a separate qusetion

eprodrom: probably unenforceable
... very difficult to test whether someone has considered something

hhalpin: testing privacy is hard
... there are ways, but they're not part of w3c test suite or well understood, basically research projects, maybe in the future
... obvious things like breaks same origin policiy we can note that
... Point on social web protocols.. seems like we're not going to converge... If we don't converge and we have this document saying there are 3 things
... And if someone has the spare time to shim between these formats
... if such a shim existed that would make SWP more useful

<rhiaro> some parts are easy to shim, some overlap completely, some not

Karli: would be desireable to have an overlap
... instead of trying to solve it like a failed attempt with a library

eprodrom: there are some formulations we could use to define it. It's an interesting question but I want to bring us back to document status

aaronpk: Beyond testing, I would like to know what is then expected of me as the editor or us as the group to move forward
... Reading the w3c process documents is overwhelming, so what is the human readable version of the next step?

<hhalpin> Note that Test the Web Forward stuff doesn't apply here

aaronpk: Assuming we close issues and build a test suite, what's left?

<hhalpin> But basically, the Process Doc is pretty easily explained and its legal to publish different documents.

eprodrom: then we have the decision of whether we're going to publish one, two or some convergence

<hhalpin> As long as they can all fulfill CR

<hhalpin> That being said, it would obviously be more desirable to have a convergence

eprodrom: We resolved to move them all forward with no dependencies

tantek: we should ask the same things as as2 of every other draft

eprodrom: test suite, conformance section

tantek: just want to be consistent

aaronpk: conformance is already in there
... oh no, micropub has it, not webmention
... okay, will do that
... if it applies to every spec can we put this list somewhere?


sandro: webmention is simple enough we can say 2 complete senders and 2 complete receivers
... exit criteria is 2 implementations of each feature

tantek: define what 'feature' means
... sandro said sending and receiving are features, but you said features break downmore

aaronpk: features like updates and deletes
... existing implementations are various overlapping subsets


cwebber2: I dont' know if we need to leave owen on there, should we remove him because he's not a current editor?

sandro: make him an author or former editor

cwebber2: we can do that
... In terms of things that are next, we have a list of actionable bugs
... Might be useful to use some call time
... Some I've been stuck
... Issues are slow
... Easier to go through on a call
... I think we'll make a lot of changes
... I'd also like to work on implementations
... So we can test the federation stuff works the way that we expect it to
... Media goblin is a good option
... Maybe somebody in the pumpio community will jump on it
... Possibly something smaller

<eprodrom_> !!!!!!!!!!!

cwebber2: Bugs, then implementations, then the whole conversation about conformance

<eprodrom_> "1.1.0 initial AS2 feed (beta)"

cwebber2: I probably will need guidence
... As for worrying about the test suite, I'd like to do that after the implementations are at a certain stage, so we know what we're testing

<ben_thatmustbeme> eprodrom_ cool!

eprodrom: work on bugs, do a new version, get review, then start talking about test suite?

cwebber2: I'd like to see implementations before test suite
... I don't think test driven makes sense for standards, when you don't know what you're pushing forward

<cwebber2> - close actionable bugs

<cwebber2> - work on implmentations (1 or ideally 2)

<cwebber2> - conformance section and security considerations

<cwebber2> - test suite

<KevinMarks> surely you will write tests as you write implementations; the question is if you can generalize them

<cwebber2> KevinMarks, quite probably, you probably heard what I just said :)

<Loqi> Tantekelik made 1 edit to Socialwg

<cwebber2> - and exit cr in between the last two

cwebber2: having a mniinmal implementation that includes tests would be good

eprodrom: date for next version of ap?
... may?

cwebber2: ...sure, mid-may


aaronpk: Add privacy and security section
... conformance is already there
... make a plan for building test suites
... That doesn't go in the spec right?

tantek: yeah

aaronpk: CR exit criteria
... And several combinations of things in micropub spec, multiple kinds of server and client implementations
... A server or a client does not have to support all features, so there are many different kinds of valid implementations
... Would like to publish another update of this before asking for it to go to cr

tantek: Also for all, identify what you consider at risk

<cwebber2> rhiaro, cool

tantek: Once you have figured out what the features are, among that set of features decide if there are any that you consider at risk
... If you think it's too new or you're not confident in it, label any of them (preferably not all) at risk and let the wg know
... people in wg, if you object to a feature file issues or ask for it to be put at risk. Or if you think a feature is essential you can argue against putting at risk
... Editors can say what you consider at risk and why
... Or you can say none ofthem are at risk
... The effect on the process is that if you get to the end of the cr period you can drop at risk features without having to publish another draft

<Arnaud> I agree with Tantek, At Risk is a powerful mechanism but I would invite anyone to make suggestions/requests

<Arnaud> this is not limited to editors

tantek: when can we expect a new draft of micropub?

aaronpk: both of them by april 8th at the latest, hopefully earlier

tantek: if you dont' meet all requirements for cr, you can still publish another wd immediately
... for all drafts
... if you're resolved issues
... close to 0

aaronpk: in that case I will propose to publish a new draft on the 29

on the next call

cwebber2: issue closing party

post type discovery

tantek: I have had issues finding the time to convert that into the proper format to publish so I need help
... I don't have a godo workflow for editing that draft on the wiki then turning that into a form that I can publish on w3c

annbass: what's involved with converting it?

tantek: if you use respec it expects a certain syntax
... which is not mediawiki


tantek: so either I have to manuall convert it every time, or convert it once and keep it in sync, or give up on the wiki and just use github
... I was hoping I could pipe the wiki page into the publication process and have it work

hhaplin: when people have edited specs in wikis in the past... it's possible..
... you just have to put effort in to convert, but it's a lot of work

tantek: I could iterate on issues and close them next
... in editors draft
... would be pgoress before trying to do conversion

eprodrom: and what form would help take?
... you want a coeditor, or general participation in wiki editing?

tantek: looking for suggestions in conversion process
... wiki is good for iterating, but not for publishing
... Open for suggestions
... If I have to convert to respec in github then fine, but I'm looking for suggestions

eprodrom: any other questions?
... If not, we are at the end of our agenda. Pretty impressed.

tantek: worth spending a couple of minutes on bringing up other docs we had as editors drafts


tantek: Action Handles
... Handlers
... was separated from as2
... only broken links

<KevinMarks> Action handlers is a broken link

<trackbot> Error finding 'handlers'. You can review and register nicknames at <>.

tantek: Not sure what happened to it. I think we have enough work, so we drop this as a work item?
... Not to say we ban it, but fo rnow we admit that nobody is actively working on it
... If someone wants to bring it back they can do that in the future
... But for now we can move it from the list of drafts
... jasnell is not here so I want to give him a week to pipe up

<Arnaud> It's ok

<Arnaud> jasnell won't oppose this motion

cwebber2: we should put it on the table and mail the list

ben_thatmust: we can resolve to remove it and bring up objections in the next telecon

<cwebber2> ah well then Arnaud :)

<cwebber2> easy!

<tantek> PROPOSED: Drop Action Handlers Editor's Draft from our list of working group drafts, without prejudice

<wilkie> +1

<cwebber2> +1

<bengo> +1

<aaronpk> +1

<ben_thatmustbeme> +1

<eprodrom_> +1

<sandro> +1


<KevinMarks> +1

<annbass> /me I like the "without prejudice" bit

<dmitriz> +1

<tsyesika> +1

RESOLUTION: Drop Action Handlers Editor's Draft from our list of working group drafts, without prejudice

tantek: the other one is jf2

ben_thatmust: I have not had much time due to personal small people reasons to work on this
... I certainly would like to see something of this go to a note, because I think it's been useful and it's standardising several implementations that are using json formats of microformats
... but I don't have the time to push that toward a rec

sandro: is it dependant on microformats?

aaronpk: I have found it useful for dealing with apis and services that use micropub and webmention, it sort of sits in th emiddle of everything and is used by both of thise, but I don't need to normatively refernece it as a spec, but ti has been useful in building implementations
... that's how I have some iterations of the document that I have been taling to ben about but haven't been updated or anything
... but it's been useful
... but I don't need to tie the specs to it
... I could if it was there, but

tantek: what do you need to do as editor to produce a working draft towards a note?

<Arnaud> sorry, I've got to go - see you

ben_thatmust: I don't know
... definitely needs editorial changes
... it's difficult because I'm definining the spec without the vocabulary. Trying to say was written with microformats in mind but you could apply any vocab to it

aaronpk: the more useful parts I"ve found are tied to the vocabulary

<KevinMarks> microformats2 parsing doesn't define a vocabulary either

aaronpk: Using as this is the one way to represent a blog post or an author with the microformats vocabulary
... Useful to me whether or not it's a spec

<KevinMarks> I did a generic mf2 to jf2 converter

tantek: if you want to take it to note we can make this intention clear up front
... so we can plan for that as a group
... non rec track draft
... When do you think it will be in good enough shape to publish as a non rec track draft

ben_thatmustbeme: I need some time to work on it
... Can say... april

<eprodrom_> sandro: So, there's a WordPress plugin that implements AS2

eprodrom: time to wrap up. Any other agenda items that we need to address before the end of the day?

<sandro> dinner is 6:15

<aaronpk> sandro++

<Loqi> sandro has 30 karma

<sandro> (dinner isn't sponsored)

eprodrom: Agenda for tomorrow
... tomorrow afternoon, can go through issues that editors want to bring to the group

aaronpk: I do have issues on wm and mp that I would like to discuss

<Loqi> Tantekelik made 1 edit to Socialwg/2016-03-16

aaronpk: Not many, like 2 on mp 4 on wm

eprodrom: I'll update agenda
... FIN

<KevinMarks> St Patricks Day in Boston?

<KevinMarks> I would think so

<bengo> night yall enjoy dinner