Socialwg/2015-07-14-minutes

14 Jul 2015

Attendees

Present: eprodrom, aaronpk, sandro, hhaplin, rhiaro, cwebber, kevinmarks, jasnell, ben_thatmustbeme, csarven
Regrets Chair: eprodrom
Scribe: rhiaro, hhalpin

jasnell: Just need the auth token. Everything else I should be able to set up on my own. If any issues come up I'll work with harry or sandro directly
... Did add more AS2 items to agenda
... Updates based on discussions on github. I have added Organization back as an Actor. There was discussion in favour.
... Removed as:Favorite as an activity type because of the overlap with Like
... Favorites are really a collection of things that have been liked
... If folks do not agree let me know

<tantek> favorites--

<Loqi> favorites has -2 karma

jasnell: Also restructured the git repo to be a closer match with what's being published. New directory structure, so the links to the editors drafts have changed, but I've added redirects
... Added more expansive friend request example

<ben_thatmustbeme> in favor of removing as:favorite (actually get that minuted)

<KevinMarks> Slack's extension of like/fav to liek with emoji is intersting

jasnell: These additionals and removals need to be reviewed

eprodrom: You removed as:Favorite not Favorites

jasnell: yes

<KevinMarks> http://slackhq.com/post/123561085920/reactions

jasnell: Typo in the wiki. Favorite not Favorites

eprodrom: you've called for specific responses. I'm interested in the change in the purpose. That seems to be closely related to API issues
... Not sure if that's something we should see come up more on Social API
... Any feedback on this? Adding purpose to collections?
... This was discussed just before the call

jasnell: I'll give an overview of the issue
... The basic idea is to add a new property to an object which identifies its semantic purpose, or to use ane xtension on the @type
... We have a purpose field that says eg. purpose:favorites, or @type Favorites
... Could exist on Collections or any other type of object to give you a context on why it would exist

<eprodrom> http://w3c-social.github.io/activitypump/#collections

<ben_thatmustbeme> which would be better for multiple types within a collection?

eprodrom: activitypump defines a number of different collections for each actor

<aaronpk> why do we need anything other than a name for the collection?

eprodrom: You would see that as something we'd add in addition, to specify what those collections are for

jasnell: Be good if peole can weigh in on github or next week

<jasnell_> https://github.com/jasnell/w3c-socialwg-activitystreams/issues/188

aaronpk: I do have what you're calling collections on my site. Different sets of things like the things I've favorited, or posts about bike rides. What is the purpose of making a machine readable value for the type?
... If you just name them, people can figure out what they are. What's the use case for machine readable?

jasnell: discoverability

aaronpk: You can look at the names of things

<ben_thatmustbeme> also, a type doesn't really give reason

jasnell: if it's a human doing the discovery then yes that works
... If you want to automate the discovery, having an identifier that's consistent and not subject to localisation issues, translations etc, then having the machine readable identifier is preferred

aaronpk: Okay, so finding specifically the list of what someone has liked, if you want to find everybody's streams of what they've liked
... Implies there is a vocabulary for this

<Loqi> Tantekelik made 1 edit to Socialwg/Social syntax/User Stories https://www.w3.org/wiki/index.php?diff=85037&oldid=81653

jasnell: It's very similar to the use of link rels, but rather than attaching it to the link it's attaching it to the object
... Just giving it a consistent label

<Zakim> tantek, you wanted to note that none of our user stories actually explicitly talk about collections, perhaps we should drop the feature from AS2

<cwebber2> +q

tantek: Wondering if the challenges we are having discussing collections is one example of... AS2 has a lot of features in that arrived with the first draft, but for example the whole notion of a collection is not actually in any of our user stories, except for one with multiple objects. User story about voting on user stories. Other than that there's no user stories with collections
... Maybe it's something we can drop.

<cwebber2> -q

eprodrom: I'm not sure that's the case. In AS, most of the ideas of a stream or having multiple activities or objects available at a particular endpoint are implemented as collections
... Like being able to navigate someone's list of friends

<cwebber2> I was just going to chime in that inboxes/outboxes in activitypump are collections

<wilkie> how can you have a stream without the concept of a collection

tantek: I guess the challenge I'm having is that collections are like a simple array. Which I agree does make sense.
... That's just a data structure

<cwebber2> so that's a lot of our user stories

tantek: But when we start talking about giving a collection a special label or type, then I start to think that's not necessary for just having lists of friends
... The term collection might be being misused, or at least confusing

<hhalpin> A collection tends to be an unordered list, while an array is an ordered list.

<wilkie> you still need some way of identifying the collection in meta. like, how many items does it have? etc

<hhalpin> i.e. a set vs. an unordered set.

ben_thatmustbeme: Couple of qustions. Once you specify one type: here's a collection that only contains favorites. What if my collection contains multiple types?

<hhalpin> I think you can call it more or less whatever you want :)

<aaronpk> i plan on publishing collections with multiple types of posts fwiw

ben_thatmustbeme: I could just to through the collection and find the type of everything that's in there

<csarven> Wouldn't different typed items belong in different collections?

eprodrom: Good question
... In case of favorites, the thing favorited might be 'audio', 'video', 'note', etc. What makese them favorite is that they're members of this collection

<tantek> this is odd because in practice we do have "likes" as objects/posts

jasnell: that's exactly the point

<aaronpk> I already have multiple post types in the same "collection" actually (what I would call feed/stream) http://aaronparecki.com/metrics

jasnell: Just knowing the type of objects that are in that set don't tell you why they're there

<eprodrom> friends

<aaronpk> here are my likes http://aaronparecki.com/likes

<melvster> collection is like an object in JS {}

<cwebber2> I do like the way favorites/likes currently show up as objects in pump.io currently

jasnell: With AP, this notion of inbox and outbox. If you have one collection URL, how do you know whether it's an inbox or outbox? How do you know what it is for?

<cwebber2> and yeah the permissions thing does happen, sometimes I see someone liked a thing I can't see what it is :)

<cwebber2> that's an interesting thing

jasnell: Having some way to explicitly identify that roles makes it much more explicit

ben_thatmustbeme: Sounds like we need another set of vocabularies for types of collections

<KevinMarks> no, melvster because {} requires named keys

jasnell: I'd rather avoid it. I'd rather not have it be extensive as another set of terms. But looking at the API we already have these notions.. We already see merging some of these concepts of different types of collections

tantek: What API for inbox?

<KevinMarks> it a list [] without assumed order

jasnell: ActivityPump, pump.io

eprodrom: There's also list of friends, followers, people you're following, list of people you've put into a collection (like g+ circles).

<ben_thatmustbeme> in many cases rel= and description could solve a lot of the issues

<melvster> eprodom++

<Loqi> eprodom has 1 karma

<cwebber2> collections are critical to activitypump's current design

eprodrom: The thing that's interesting here is that instead of having a list of friends, we have a list of the activities that created those friends

<melvster> eprodrom++

<Loqi> eprodrom has 22 karma

eprodrom: I think that's an interesting question. I wonder if there is value to one side or the other
... To me I'd rather have the people than the activities

<cwebber2> it would be a total restructuring of everything if that were dropped

<ben_thatmustbeme> +1 to list of People, not activities

eprodrom: I don't know if there's a good answer for the other

<cwebber2> and it would have serious repercussions in what tsyesika is doing structuring things in mediagoblin too

jasnell: For things like your friends list, we have the relationship object, so you could have a collection of thos eobjects. You could infer friends from that. So not necessarily have the explicity purpose
... But for things like likes, favorites, sharing, saving. Those are much more difficult to infer

<cwebber2> lots of things, playlists of audio/video, feeds of things you like, feeds you subscribe to, those are pretty critical

(infer from properties?)

<tantek> or from name?

<eprodrom> example.org/evan/favorites

eprodrom: Another thing you cold do to infer it is look at /favorites and see if a collection is returned to me
... Then I know what that is
... /favorites is an example url, could be /whatever. Just because I was looking for favorites.

<tsyesika> cwebber2++

<Loqi> cwebber2 has 38 karma

eprodrom: Seems useful, not sure if it's crucial

tantek: I think I understand now what James was saying about rel values
... If you have a rel value that says rel="likes-feed" that gives you a way to do discoverability and the feed itself desn't have to say anything

eprodrom: this is almost the rev in html temrs
... These are the likes of someone else

<KevinMarks> these are others' likes of my post? a facepile?

jasnell: There's a number of different ways of doing. Need to make a decision. Do we want to attempt to do everything implicitly where we look at the contents of the collection to see what it's for or do we want an explicit identifier
... Or do we do it with the object itself using link rel
... Just need to know what people would like to see
... Issue 188

eprodrom: Let's contineu the conversation in the issue
... And see if we have an update for our next call
... Let's move on if no more questions

Social API

There aren't actually any updates

<hhalpin> scribenick: hhalpin

<rhiaro> https://github.com/w3c-social/Social-APIs-Brainstorming/blob/master/socialapi.md

<rhiaro> Not sure if people have looked at that

Should we move onto the Webmention topic then?

<rhiaro> Just a few headers, with some comparison stuff

<rhiaro> Not changed since last week though

<melvster> I updated: https://github.com/w3c-social/social-ucr a fair amount

<eprodrom> rhiaro: are you on the call?

rhiaro: we've put this document together but haven't had much feedback
... no time to update, so it's the same
... however, we encourage people to look at it, will be working on it by end of week
... start of strawman

eprodrom: moving from comparison to strawman
... still micropub and activitypump evolving side-by-side
... we'll continue this development and this doc will track those

rhiaro: if there's a convergence we add it there

eprodrom: Next steps is continued work in activitypump, micropub, and tracking similarities

tantek: As far as an update is concerned, I would be intereted in hearing when the prototypes are done
... there is a certain status right now, but how much is prototype and kinda working?
... or how much is an API design that hasn't been prototyped?

aaronpk: My goal is to drive contributions while I have code working on it
... I've got a backlog of code
... so that's why I haven't contributed as much
... as I'm coding first

eprodrom: Everyone on the call should understand that this work will have a convergened version that will have one spec, correct?

aaronpk: That was my understanding, but I don't want to make suggestions till the code works

rhiaro: Mine as well
... comparisons are definitly potential, as well as where htings are done diferently, maybe two optioins in one spec

eprodrom: Very reasonable sounds good!

charter license update

<rhiaro> scribenick: rhiaro

sandro: Nothing new to share with the group. Issues to straighten out, hoping to get approval from w3c tomorrow or a week from tomorrow, then will go to AC

<hhalpin> scribenick: rhiaro

tantek: One of the actions assigned to me is define how implied post typing works. The more I write that up, the more it sounds like a spec
... Follow a sequence of steps. If I"m goign to contribute that as a spec, I'd like to do that with the new license
... So hopefully we'll get through the rechartering quickly

sandro: understood

eprodrom: Moving forward..

SWAT0 implementation

tantek: As background, SWAT0 is the Social Web Acid Test, is something that myself, evan and david (??) came up with almost five years ago at the Federated Social Web Summit

<KevinMarks> recordon

tantek: to provide a difficult challenge for different integrated social services to demonstrate that they are federating in such a way that satisfies a fairly common use case, even at the time, that people were using social networks for
... I'll summarise
... A takes photo of B with their phone and posts and tags B i in it
... Then B is notified that someone posted a tagged photo

<KevinMarks> SWAT0 A posts+tags mobile photo of B B photo notified C(follows A) sees it; replies A&B comment notified

tantek: Person C who is following A, sees the photo post from A and in the same interface, replies or comments o nthe photo
... Both A (author) and B (tagged) get notified
... about the comment
... THat's a fairly common thing people do on any photo social network service
... But noone does it across services

<melvster> FYI: https://github.com/aaronpk/Micropub/blob/master/user-stories/SWAT0.md

<hhalpin> +1 for getting this working!!

<aaronpk> ooh I need to review that to see if it's accurate now that it's been implemented

tantek: This weekend, we got that working with three different people, using three different pieces of software, running on their own sites, working with mobile phones
... A was ben_thatmustbeme
... took ap hoto with his phone of B (aaronpk)
... Posted it to site from his phone with micropub

<KevinMarks> details with links: http://tantek.com/2015/193/t2/user-flow-permalinks-indieweb-swat0

tantek: used webmentions to notify aaron's site that he tagged him in a photo
... aaron got notification on his phone immediately
... By the way, 3 different locations
... Portland, Boston, SF
... kylewm, subscribed to ben, saw the photo and commented
... comment went to ben and aaron
... ben and aaron got notifications
... There's a writeup and a video
... Appears we got that working
... This is our one approved user story
... Which we got working with micropub, webmentions and microformats

eprodrom: Great achievement. Congratulations in order

<melvster> link to video would be great ....

<ben_thatmustbeme> didn't we approve the unanimously +1'd user stories too? need to check previous minutes to double check

eprodrom: Good show of the continued maturing of micropub, webmention and microformats
... Also an interesting challenge to folks working on SoLiD and ActivityPump to show the same kind of activity

ben_thatmustbeme: didn't we approve the +1'd user stories?

tantek: I vaguely remember that, but couldn't find it on the wiki. Can someone find that resolution?
... We should udpate the approved user stories accordingly
... Anybody else?

eprodrom: I'll take an action to see if we have that

<tantek> here's what we have so far: https://www.w3.org/wiki/Socialwg/Social_API/User_stories#Approved_user_stories

eprodrom: If we do I'll update the approved user stories

<melvster> I have 13 listed here ... SoLiD have 5 implementations currently ... 6th in progress : https://github.com/w3c-social/social-ucr

eprodrom: I remember it too

tantek: maybe one of us just dropped the ball in updating the wiki
... The nice thing about SWAT0 is that no one person or implementation can claim to implement SWAT0. You have to have at least two different pieces of software
... One of the things we learned that there are very different implementation requirements for A vs B vs C
... Identifying those requirements was part of the challenge
... Especially for people making their own, it may be easier to play on eof those roles rather than another
... But then, it maybe be possible to create software that does all three
... There aren't any I'm aware of currently in indieweb
... statusnet originally could do all three?
... That's maybe another bar to achieve. A piece of software that can play all three roles. I don't think we've had any since statusnet

eprodrom: This is good step forward. I appreciate everyone's work on it
... Anything else to talk about here?

tantek: Any questions?

<melvster> would love to see the indieweb version of SWAT0 interoperate with one of SoLiD and/or ActivityPump, if that is at all feasible

<eprodrom> melvster, are you on the call?

<melvster> following on irc

eprodrom: I'mnot sure if that is feasible. By the definition of those groupings.

<KevinMarks> if Solid marks up with mf2 and sends webmentions, sure it is

eprodrom: We're not talking about pieces of software as much as groups of software that implement particular protocols

<aaronpk> you could certainly upload the photo via a SoLiD api instead of the micropub api

tantek: certainly possible if there's a solid or AP implementation that published microformats or sends webmentions. Would be a cool bridge.

<KevinMarks> you cna send webmentions externally to the page too

<KevinMarks> so just needs mf2 in html

tantek: If anyone is implementing SoLiD or AP based pieces of software, see if you can also publish in your html microformats
... If you need help with webmention there are a ton of libraries that are simple to use to call that

<KevinMarks> (I think tantek and me are saying the same thing in parallel)

<aaronpk> check out http://indiewebcamp.com/SWAT0#implementation_requirements and see which of A, B or C is easiest to implement

eprodrom: The client API could be solid or AP or micropub, and the server to server is using webmention

<KevinMarks> exactly evan

tantek: each of the different players is talking whatever with its server
... the federation depends on webmention and microformats as far as I can tell

<KevinMarks> also the notification wiring

eprodrom: yes. Interesting to see that happen

<ben_thatmustbeme> that would be cool

eprodrom: Move on:

Outstanding actions and issues

eprodrom: No pending or raised issues that we haven't dealt with

<eprodrom> http://www.w3.org/Social/track/actions/open

eprodrom: Would like to ask if we have any of our open issues that we'd like to report progress on?

<eprodrom> http://www.w3.org/Social/track/actions/71

eprodrom: I will report progress on action 71

action-71

<trackbot> action-71 -- Evan Prodromou to Break up github issue 133 into several ones, and close 133 -- due 2015-07-14 -- OPEN

<trackbot> http://www.w3.org/Social/track/actions/71

scribe: Is now complete. Will close.
... Any others?
... So that's the end of the agenda.
... Anything else?

<ben_thatmustbeme> https://www.w3.org/wiki/Socialwg/2015-05-12-minutes

<ben_thatmustbeme> ' RESOLVED: accept all user stories that are entirely positive as official user stories '

eprodrom: One thing that tantek, I and staff contacts have discussed is cutting down on our meeting schedule for the next few weeks, since we are into vacation season

<tantek> ben_thatmustbeme++

<Loqi> ben_thatmustbeme has 100 karma

<KevinMarks> well found ben

<cwebber2> and I'll be at OSCON next week

<cwebber2> (will anyone else?)

eprodrom: Starting next week, we would skeip every other meeting
... We would have no meeting on 21st, but one on 28th
... (etc)

<tantek> ben_thatmustbeme: FYI, "entirely positive" means these: https://www.w3.org/wiki/Socialwg/Social_API/Sorting_user_stories#Entirely_Positive

eprodrom: In september we would pick up our regular meeting schedule
... To deal with absences etc
... any objections?

<aaronpk> cwebber2: I'm in Portland but not going to oscon!

tantek: so next meeting 28th?

<wilkie> cwebber2: I'll be in Portland too!

eprodrom: yes

<wilkie> aaronpk: let's have a get-together

<aaronpk> video of swat0 demo! http://aaronparecki.com/notes/2015/07/14/1/indiewebcamp

<cwebber2> aaronpk: wilkie: ooh, yeah get-together would be great

eprodrom: ajourn!

<eprodrom> trackbot, end meeting

<cwebber2> ok later everyone!