From W3C Wiki
Jump to: navigation, search

Social Web Working Group 3rd face-to-face

04 May 2015

See also: IRC log


INRIA, ben_thatmustbeme, cwebber2, eprodrom, Elf_Pavlik, AnnB, Olivier_Berger, Arnaud_LeHors, Wendy_Seltzer, Melvin_Carvalho, Andrei_Sambra, Jessica_Tallon, Amy_Guy, Aaron_Parecki, Sandro_Hawke, Paul_Tran-Van, Harry_Halpin


Summary of Action Items

[NEW] ACTION: pelf to draw Follow vs. Subscribe with account having multiple feeds allowing subscription independently [recorded in]

<Arnaud> trackbot, start meeting

<trackbot> Date: 04 May 2015

<sandro> James, tell INRIA bridge the conference code: 23695487

<sandro> hoping this works.

<sandro> jasnell, does that make sense? We can hang up and call it again

<sandro> jasnell, using "Zakim, drop INRIA-bridge" or "zakim, call INRIA-bridge"

<jasnell> sorry, was off getting coffee

<Arnaud> we're getting zakim to call the local bridge but zakim won't enter the passcode

<jasnell> sandro: sorry, no, not sure what you're asking me to do

<Arnaud> if you hear the local bridge asking for the code, try to enter it yourself

<jasnell> not hearing anything

<Arnaud> the conference code: 23695487

<Arnaud> did you hear anything?

<jasnell> yes, trying

<jasnell> need more coffee

<jasnell> it will be difficult for me to hear and understand everyone. I will be relying quite a bit on whoever is taking minutes

<Jessica_Lily> I'm Jessica Tallon (Tsyesika) from GNU Mediagoblin

<sandro> Sandro Hawke, MIT & W3C

<AnnB> Ann Bassetti, Boeing

<rhiaro_> Amy Guy, University of Edinburgh

elf Pavlik - independent / polyaffiliated

<AnnB> scribenick: AnnB

<wseltzer> Wendy_Seltzer, W3C

<aaronpk> Aaron Parecki - indiewebcamp

Arnaud: will work on: Activity Streams, make progress on protocol, ...

<jasnell> *trying*

<elf-pavlik> jasnell++ for staying up at night to join us :)

Arnaud: change from original agenda, to have James Snell (@jasnell) on first, to accommodate Pacific time zone (middle of night)

<Loqi> jasnell has 13 karma


<Loqi> jasnell has 14 karma

I would appreciate if the rest of you could put your name and association into IRC

so we have the proper list of attendees

James Snell: number of issues in Tracker


re: Activity Streams...
... have 1st iteration of test harness .. loaded on Saturday

<melvster1_> name : Melvin Carvalho (independent)


... node-based application
... need to start iterating on it
... please help
... this is the test harness jasnell's colleague JP demo'd
... jasnell is busy for next couple weeks, but then will work on it
... good beginning for test harness

Arnaud: (clarifying).. we pushed JP to put this harness out there, even though the code wasn't done
... it's not his job, we "volunteered" him
... but we wanted to get it out there, so people can start helping
... creating tests
... would be really helpful if WG members would let us know who can work on this

Sandro: could someone give an overview of how this harness works?

Arnaud / jasnell: JP is in process of documenting

jasnell: give an AS instance
... it will check if it's valid JSON

<Arnaud> JP=Jacques Perrault

<then other things happen that I missed ... maybe @jasnell can type later>

Sandro: it runs as a web service?


jasnell: yes .. runs as a node web app
... goal is to be able to install locally, but not quite ready on that

sandro: why?

jasnell: to be able to test your own docs, if you don't want to share publicly

<jasnell> Ann, I will be documenting the fundamental test process

arnaud: right now it's rudimentary, but it's only beginning

<jasnell> I'll try to get that written up later in the afternoon today... after I get sleep

arnaud: goal is to demonstrate we have interoperability in the spec

<elf-pavlik> Arnaud: as WG we need to deliver implementations report

<Zakim> elf-pavlik, you wanted to discuss how people implementing libs and tools can use it for automated testing?


elf: I see there are 2 implementations started
... are there plans to automate?
... for people who built libraries
... another guy I will meet with in Dusseldorf has some stuff
... how can he work on this

<elf-pavlik> Nick (working on sockethub) writes different JS AS2.0 library

jasnell: that's the goal, but I'm not clear exactly on how it will work

<Zakim> sandro, you wanted to say So it's a Feed Validator, at the moment, not a Test Harness

jasnell: ideally we'll be able to drop into some environment, and it'll just run... but none of that is worked out yet

sandro: what threw me off is the name "test harness"
... what you really mean is "feed validator", right?

jasnell: yes, right now only validator ..
... longer goal is more

arnaud: what is expected?

sandro: nice to have test feeds and something to do with them

arnaud: that's part of the challenge, there's only so much you can do in this case

sandro: I guess you could have some valid and invalid AS

jasnell: yes ... will come more into play when we have an API

sandro: there

s/sandro: there //

<jasnell> there needs to be a complete corpus of test documents, both valid and invalid

<Zakim> elf-pavlik, you wanted to discuss testing expected *side effects* after adding an Activity to dataset


<sandro> sandro: So (1) Feed Validator, (2) Syntactic Test Suite -- wiith hundreds of Good and Bad documents, for people to make sure their parsers are good

elf-pavlik: we were talking about testing side effects
... how do side effects happen ..

<jasnell> I am sorry but it's very difficult for me to understand what elf is saying

elf-pavlik: before and after activity happens

<sandro> elf-pavlik: For the client API, we can test the side effects, eg what a subscription changed

arnaud: we can only validate the stream

sandro: once we know what API we're doing, we can figure out better how to test it

elf-pavlik: can test what it looks like before activity and after activity

jasnell: we can test "stories", based on our user stories, for certain scenarios
... what is expected output
... with some syntax variances

<Loqi> Wseltzer made 1 edit to Socialwg/2015-05-04

jasnell: for now we can only go so far

elf-pavlik: I'll try to make a pull request on what I was suggesting


elf: trying to illustrate; we can see if it works or not

<elf-pavlik> i'll make pull request with test for how Follow activity affects the dataset

arnaud: anything else on test suite?


jasnell: moving on with open issues
... there are 13 open issues
... Tracker issues (not github)


jasnell: 2 new ones

<wseltzer> ISSUE-20?

jasnell: 1 by Eric Wilde

<trackbot> ISSUE-20 -- Represent Collections using JSON Text Sequences (RFC 7464) -- raised


jasnell: don't want to close it without Eric being able to discuss it
... would rather wait for concrete proposal

<elf-pavlik> -1 to fiddle with tracker now

jasnell: therefore recommend we do not open it


<elf-pavlik> can we discuss

<elf-pavlik> issue-36

<trackbot> issue-36 -- Role and evolution of recommended JSON-LD contexts -- open


<hope jasnell can type his explanation into IRC later>

... not clear how this would work with JSON-LD, nor what it would "buy" us
... we currently represent as a document with an items array

elf: for most people here, this is hard to discuss ...
... might be better use of our time to discuss items we are engaged with

<jasnell> AnnB: let me know if I'm speaking too quickly or not clearly enough.

arnaud: this isn't "random"; jasnell is the editor, and he's just going through the issues

<Zakim> elf-pavlik, you wanted to discuss picking only issues most relevant to us dissing them F2F here

elf: .... that's just my observation; I might be wrong

<Zakim> sandro, you wanted to ask is this precluded in any way, via con-neg

jasnell: yes there's a proposal, but no examples on how it should work

sandro: this sounds like something that could be added later based on context negotiation, yes?

jasnell: yes
... agree

arnaud: the proposal is to close this issue (that is, not open it)

jasnell: if someone can come with a concrete proposal later, they are invited to raise it again

<Arnaud> PROPOSED: close issue-20 until someone comes with a real problem and concrete proposal

<sandro> +1

<elf-pavlik> 0

<jasnell> +1


<deiu> 0

(leavingn it to the geeks)

<wseltzer> +1

<obergix> (just an observer)

<rhiaro_> 0

<Jessica_Lily> +1

<melvster1_> 0

<Arnaud> RESOLVED: close issue-20 until someone comes with a real problem and concrete proposal

<parklize> 0

<elf-pavlik> issue-36

<trackbot> issue-36 -- Role and evolution of recommended JSON-LD contexts -- open


jasnell: can we talk about raised issues first?


jasnell: we don't need to solve this one yet, but I do recommend opening it

sandro: we can't make progress on this until we decide on API
... but probably OK to open it

arnaud: I find this a bit premature
... LDP stuck

<Arnaud> PROPOSED: open ISSUE-37

<elf-pavlik> +1

jasnell: point of reference: I've implemented it both ways
... found that clients implement both ways

<sandro> sandro: This makes sense as an issue with or without LDP. It's the basic question of whether to have paging links in LINK headers or inside the data. Good question.

jasnell: I think it'd be good to at least leave open

Harry arrives

<aaronpk> +1

<sandro> +1

<jasnell> +1

<Arnaud> RESOLVED: open ISSUE-37

jasnell: OK, now to the open issues
... quite a few of them


elf wants to start with #36

elf: I find it interesting that we try to have JSON and RDF
... how we design and implement will be crucial
... will it be frozen or can it still adjust over time?
... lately jasnell was flattening namespaces ... which makes it hard in RDF
... how do we handle this
... how do we understand the JSON context?

<tantek> ASIDE: Was there any discussion in the f2f meeting about explicit goals for the meeting? So far there is only "Proposed Goals":

<tantek> Goals currently are "TBD"

jasnell: re: context ... keep in mind: 1) there is a context that is normative; 2) keep it minimal


<tantek> Arnaud: ^^^

jasnell: I think we should remove a lot of the non-AS vocabulary other than vcard

<Arnaud> tantek: not much of a discussion I said the goal for today is to close as many issues against AS as possible

<can someone else fill in this bit?>

<tantek> Thanks Arnaud - will update accordingly.

<Arnaud> thanks

<Loqi> Tantekelik made 1 edit to Socialwg/2015-05-04

<elf-pavlik> sandro if term 'foo' doesn't have URI will get blank node _:foo

jasnell: challenge is json-ld contexts are not perfect ... no versioning
... keep context as minimal as possible

<elf-pavlik> if it dosn't map in context to URI

<elf-pavlik> sandro please see

arnaud: asks sandro what the policy is in publishing this kind of stuff

<jasnell> json-ld contexts are not versioned, there is no way to track backwards incompatible changes across json-ld context versions

sandro: namespaces won't break interop; but contexts will

<jasnell> if we keep the json-ld context as minimal as possible, implementations can extend as necessary without us having to do much

sandro: in new Process doc, it's more lightweight

<jasnell> the normative context document needs to have JUST the Activity Streams stuff + the denamespaced VCard properties

<jasnell> (in my opinion)

Harry: in order to make an change like this, we don't have to re-open WG

<jasnell> it would be extremely helpful if JSON-LD had minimal versioning support

Harry: something like "re-open" but not "re-charter"... not sure

<jasnell> {"@context": {"@version": 1.0.0, ... }

Harry: more lightweight

arnaud: does this mean we have to have it in the spec as well?

<elf-pavlik> issue-16

<trackbot> issue-16 -- better separate grammar/vocabulary and improved spec structure -- open


arnaud: what would it take to close this issue?

sandro: there are a couple aspects to this
... depends on what type of change
... there are forward and backward-compatible changes
... but changing mapping to existing namespace would break existing stuff

<Zakim> sandro, you wanted to ask if extensibility is a requirement for this WG

jasnell: it's hard to make changes that will break nothing

<tantek> yikes about versioning and extensibility - would strongly prefer forward-compat design practices

sandro: how to do extensibility? is that a solid requirement?

<jasnell> tantek: json-ld contexts are not versioned

sandro: is it important that someone be able to add an enterprise-specific XYZ?

<eprodrom> Hello all

jasnell: absolutely yes
... have real-world examples

<elf-pavlik> extensibility also relates to issue-16 !!!

jasnell: people can extend, so long as it's not part of AS namespace

arnaud: then there's no way for us to guarantee we won't break those extensions

jasnell: diff extensions might conflict .. yes .. that's not the issue .. we're talking about the core

<elf-pavlik> issue-36 says multiple contexts <- pluaral

jasnell: can add stuff via json-ld

sandro: but doesn't that break it for others, who aren't using json?
... what if one company adds an extension
... then others in their sphere also use that

arnaud: only way to get unique names is to use context (?)

<jasnell> if an implementation depends on extensions at all, there is a risk of reduced interop, that's just the nature of decentralized extensibility

harry: in any decentralized system, you can't guarantee that everyone will understand everything
... XML has same issue

elf: it's a challenge, but maybe not a problem

sandro: give warning: if you get stuff you don't understand, then what to do
... in practice json-ld will have lots of extensions

<Zakim> elf-pavlik, you wanted to discuss multiple contexts and relations to issue-16

<elf-pavlik> issue-16

<trackbot> issue-16 -- better separate grammar/vocabulary and improved spec structure -- open


sandro: harmful to have only json 'reading' of this

<sandro> sandro: Basically we need a huge warning on JSON-ONLY usage of Activity Streams.

elf: need to have diff strategy

<eprodrom> Is there not a video console set up in the room?

eprodrom: no

<eprodrom> I'm not seeing anyone else on

<eprodrom> Oh

hold on... maybe we can get it set up

<eprodrom> OK

<sandro> no, we can't. use zakim.

oh, sorry

<eprodrom> >:(

<elf-pavlik> talky will not work, we have ports blocked here

<jasnell> eprodrom: call the regular conf line

<eprodrom> Gotcha

problem is that this network blocks it


sandro: json extensibility won't work
... or, it works, but with lots of 'danger'
... ditto RDF
... we should not change the context content

elf: that's why we should be careful what we put in it

melvster: this namespace problem not specific to AS
... exist in java .. hasn't been that much of a problem

jasnell: we have already built in the <something> model ... which helps namespace your properties
... any system that has extensions runs risk of reducing interop
... best we can say is, "ignore what you don't understand"
... keep the core stable

<sandro> Not just STABLE but IMMUTABLE

<sandro> Right?

jasnell: add stuff that you want, but realize you run the risk of instability

eprodrom: AS is really quite complete

<elf-pavlik> current mapping of "Home"

eprodrom: most is there

<elf-pavlik> i bet most people don't understand semantics vcard:Homa has!

eprodrom: if you want to use a new vocabulary .. that's OK, but just know that you are extending core vocab
... most of what we need is there already

harry: I agree with Evan; plus vcard is very complete
... in current spec, would it be helpful to put something like "must ignore"?



<jasnell> "In Activity Streams 2.0, an "extension" is any property not defined by the Activity Vocabulary. Consuming implementations that encounter unfamiliar extensions must not stop processing or signal an error and must continue processing the items as if those properties were not present. Support for specific extensions can vary across implementations and no normative processing model for extensions is provided."


harry: what's current status of spec on this?

<hhalpin> should we say MUST ignore

<hhalpin> or have some specified error processing (which is what most Web APIs do?)

arnaud: I think "ignore" is too strong
... what you don't want application to balk

jasnell: want to be careful to not drop data on floor by mistake

<hhalpin> So should we say 'don't throw an error, but keep data'?

jasnell: better to say 'don't drop data if you don't understand it' ...

<hhalpin> I think that's the real question.

<hhalpin> I'm OK with keeping the text the way its written, but I just am trying to clarify what I think the concern is from Sandro and a way to address.

sandro: is there any reason we need to have only one context?

<eprodrom> Yes

<hhalpin> Historically, I'm pretty sure people don't use namespaces

<hhalpin> so I'd prefer to keep a pretty large base context

<eprodrom> +1

jasnell: if we think vcard will cause conflicts, then we could remove it
... some terms in vcard vocab need to be slightly re-worked (?)
... one wayy to close this issue is to say only AS is normative

<eprodrom> -1

jasnell: we could have 1 normative, with others as "extended"

<Zakim> elf-pavlik, you wanted to discuss distinction between AS2.0 Vocabulary and normative JSON-LD context

sandro: e.g., "AS 1 context", then "AS 2 context" ..

elf: haven't heard how we handle media types

<sandro> so lets change the AS namespace to include "2.0"

elf: json context is diffferent from vocabulary

eprodrom: disagree with having various contexts
... we should provide 1) core; 2) vocab
... if others want to add extensions... they can do that, but should not count on interop
... esp want to ensure our social API has a complete package .. then others can use extensions to add new functionalityy

arnaud: seeking proposals to close this issue

jasnell: question to eprodrom ... should we keep vcard or not?

<elf-pavlik> my proposal: create Task Force to resolve issue-16 and issue-36

eprodrom: we should have AS vocab, but make sure it's complete .. if we need to add vcard, I'm open; but let's not have multiple implementations

<jasnell> PROPOSAL: Remove all non AS2 namespaces from the normative context, keep the normative context limited to only AS2 vocabulary terms

<kaepora> Hello, I'm Nadim Kobeissi, PhD student here at INRIA :-)

thanks Nadim

<elf-pavlik> Manu Sporny uses v1 / v2 in context URIs

<elf-pavlik> from

<Arnaud> PROPOSED: Close ISSUE-36, Remove all non AS2 namespaces from the normative context, keep the normative context limited to only AS2 vocabulary terms

<jasnell> +1

<eprodrom> +1

<elf-pavlik> -1 I see need for Task Force to resolve issue-36 and issue-16 together

<deiu> +1

<Tsyesika> +1

<aaronpk> +1

<parklize> +1

arnaud: elf: explain why voted -1

elf: although we have good intentions to move on.. I anticipate problems in future from this
... would rather have small team explore this more deeply

arnaud: I appreciate that ... I prefer to close it now .. and then, later, if we find we made a mistake, then we can re-open it

<jasnell> I'm -1 on keeping issues open simply because we *might* have issues at some indeterminate point in the future

arnaud: would that be OK?

<sandro> +1

<elf-pavlik> -0.5 i don't want to block it

<Arnaud> RESOLVED: Close ISSUE-36, Remove all non AS2 namespaces from the normative context, keep the normative context limited to only AS2 vocabulary terms

<obergix> obergix is Olivier Berger from Institut Mines Telecom / Telecom SudParis, and just an observer

<Tsyesika> morning cwebber2!

<jasnell> Well, I can help us out a little bit by closing issues 28 and 32... those were previously resolved

<Tsyesika> cwebber2: we're working on AS 2.0 today and the specs tomorrow

<obergix> sandro, did you mention a proposal bout numbering a URI

<obergix> ?

<kaepora> I work here, if you guys need a place for lunch I can help you find a place

<taking short break>

<eprodrom> cwebber2: hey

<kaepora> wseltzer: It would be helpful to know of the group's dietary restrictions

<Tsyesika> cwebber2: we're taking 10 minutes

<sandro> ISSUE: Do we need to add a version number to the AS Context URI, to avoid breaking software when new terms are added?

<trackbot> Created ISSUE-38 - Do we need to add a version number to the as context uri, to avoid breaking software when new terms are added?. Please complete additional details at <>.

<cwebber2> hi eprodrom

<cwebber2> Tsyesika: aha cool

<cwebber2> Tsyesika: I didn't see anything on talky

<aaronpk> we have a ridiculously restricted network here, so no talky

<cwebber2> ahhh

<jasnell> cwebber2: talky doesn't work, we're on the regular zakim conference line

<Tsyesika> cwebber2: we're resuming

<obergix> hi cwebber2 !

jasnell: issue 13

<wseltzer> issue-13?

<trackbot> issue-13 -- Which activity types are built into AS2, and how are they defined/structured? -- open


<sandro> +1 elf-pavlik: prioritized issues!

<cwebber2> hey all, ok

<cwebber2> I was reading backscroll :)

jasnell: ones in there now .. were in original AS schema .. and then augmented
... need to reconcile against user stories, to make sure we have correct set of activities
... how do we refine that list, to make sure we have correct set?
... I think that's what Erik was getting at

<melvster1_> user stories:

arnaud: I agree; this is an open-ended question
... might be better to have several specific issues about specific activities
... therefore I suggest we close it

<eprodrom> +1

jasnell: agree

<jasnell> +1

<akuckartz> +1


<Tsyesika> +1

<aaronpk> +1

<elf-pavlik> -0.5 having doubts that we all understand this issue the same way

<Arnaud> PROPOSED: Close ISSUE-13, as is. If there is a problem with specific types missing or that shouldn't be there this should be handled with specific issues

<deiu> +1

<rhiaro_> +1

<elf-pavlik> -0 having doubts that we all understand this issue the same way

<Arnaud> RESOLVED: Close ISSUE-13, as is. If there is a problem with specific types missing or that shouldn't be there this should be handled with specific issues

arnaud: when we create issues, make them very specific
... makes it easier to deal witth them

<cwebber2> that's me

arnaud: if you can take a guess at the solution, all the better

jasnell: issue 15

<elf-pavlik> I also didn't look at issue-13 for long time so have hard time to get it from hard disk to my RAM ...

<deiu> issue-15?

<trackbot> issue-15 -- AS2.0 Vocabulary in many ways duplicates and efforts -- open


elf: explains why he thought this was a problem
... worried that it will be hard to change the vocabulary later

<kaepora> Where can I see the current draft for the vocab/context? :-)



<eprodrom> jasnell: thanks

elf: need easy way to add terms

jasnell: nothing arbitrary about current set

<eprodrom> deiu, thanks, we can go over them during the discussion tomorrow

jasnell: selected from existing social platforms
... documented in wiki
... we do need to refine these to make sure they fit our user stories
... can use existing extensibility models
... we need to define minimum set
... not interested to define all of cases ..
... but yes there will be some overlap
... not a prob
... need to define minmal set for AS2

<jasnell> my proposal would be to close issue 15

AnnB: what's your reaction, elf?

elf: current model says there's a way to extend ..
... but we don't have a way for others to use the extensions

<jasnell> that's the nature of extensions. if implementations depend on extensions, then there's a risk of decreased interop

elf: how would someone else know about new terms?

<jasnell> implementations need to agree on what extensions to use

elf: technical way to share new terms, but not
... 'social way' to share
... I suggest the IG could work on this problem

<jasnell> if the interest group wants to work on a generalized way of handling extensions in a reliable way, then fine, but this is not an issue for the AS2.0 spec itself

arnaud: but, about the spec .. you raised this issue re: the spec

<jasnell> overlap is fine

<jasnell> it doesn't matter if there's overlap in the vocabularies

arnaud: when I read this issue, it's kind of broad ... there are always these problems with vocabularies
... how is this different?

elf: be prepared to use existing terms, not duplicate them

<jasnell> the best we can do is focus on our own use cases, we cannot anticipate what others might do in the future

<eprodrom> Sorry, phone problem

<eprodrom> I'll type quickly

<eprodrom> I think we need to provide enough vocabulary

<eprodrom> That developers can build social applications with AS 2.0

<eprodrom> Especially ones that meet our user stories

<kaepora> The structure of AS2 lends itself well to the inclusion of integrity and authenticity protection features. I would love to discuss these possibilities! (I work on crypto stuff)

<eprodrom> A smorgasbord approach will not serve developers or users

<cwebber2> I would prefer a universe where we could all just use as a base vocabulary but governance is the main concern

arnaud: we agree we need to keep spec as minimal as possible, but large enough to support the use stories we've identified

elf: e.g., I understand there is some conflict between IndieWeb using their terms, others using AS terms, other using other...

aaronpk: I personally only rarely add new vocab terms
... I try to avoid adding new ones until I'm sure I need them

<jasnell> I'm not concerned in the least about overlap.

elf: if there's overlap between terms in diff vocabs .. what do you do

aaronpk: not sure

<eprodrom> Can we come up with a proposal here?

elf: we have 1); 2) AS; 3) IndieWeb / microformats ... but not coordinated

arnaud: I think elf asks we establish a general policy how to handle this

<elf-pavlik> i don't mind focusing on rather than

jasnell: I believe we've previously decided that we will not use as a normative requirement
... no spec that maps microformats to json-ld

<elf-pavlik> early draft:

jasnell: nor mapping of microfomats to activity models

<cwebber2> could we get the microformats community to bless a RDF model? :)

<elf-pavlik> tantek ^

jasnell: without anything that describes how to reliably use microformats with json-ld, I don't see how we can use it

scribe is not sure jasnell said "use it"

<Zakim> sandro, you wanted to ask if anyone cares about interop between iwc and as ?

scribe: this seems like a vague "what if" issue, rather than concrete

sandro: I hear, aaronpk, that you / IndieWeb folks aren't focused on interoperability

aaronpk: I wouldn't say that



rhiaro: I've been looking at how to model my posts with and without activities ...
... we've agreed to drop verbs

<elf-pavlik> rhiaro++

<Loqi> rhiaro has 68 karma

<jasnell> Webactions != Activities, Webactions == potential actions

<elf-pavlik> but they use the same *verbs* ?

<elf-pavlik> or Activity/Action types?

<elf-pavlik> Like / like

<elf-pavlik> Share / Repost share / repost

rhiaro: I'm using "Posts" ... similar to verbs or activities

<cwebber2> I think a translator would be great!

rhiaro: I think we're talking about similar concepts, but using different terms

<cwebber2> my main concern, and I don't mean to sound snarky: if we get an RDF model of microformats stuff, can we get the microformats community to play along, or will there be a wiki page saying "here's how to use microformats with RDF, but here's links to why RDF is terrible" :)

<jasnell> in AS2, everything is an "Object", even Activities. In IndieWeb, everything is a Post

rhiaro: I think there's a mapping, and I'm trying to think about that
... so far I have not come across anything that is not a "post"

sandro: is it correct to refer to microformats or IndieWebCamp?

aaronpk: microformat is a vocabulary and a context, and IndieWebCamp is a community

<cwebber2> rhiaro_++

<Loqi> rhiaro_ has 69 karma

<cwebber2> about trying to merge worlds there :)

<deiu> rhiaro++

<Loqi> rhiaro has 70 karma

<elf-pavlik> issue-15

<trackbot> issue-15 -- AS2.0 Vocabulary in many ways duplicates and efforts -- open


<cwebber2> bridge worlds!

arnaud: rhiaro appears in a unique position, in that she's actually been trying to map these systems against each other
... so, what's your opinion rhiaro?

sandro: I don't care about duplication, I do care about interop

rhiaro: biggest problem is that the people see the basic models as different
... but I think, in the end, they are more similar than people than think
... but I don't know what the solution is

<cwebber2> audio is breaking up here for me

<sandro> rhiaro: biggest problem is the fundamental models underlying AS2 and MF are different

<cwebber2> though AnnB's transcribing is helping!

rhiaro: other than try to implement and see what happens
... I'm trying now to post something in microformats structure
... next will try AS

jasnell: IF someone wants to create a mapping; I'm all for a concrete proposal for that mapping
... Issue 15 is much too generic, and is not actionable

<eprodrom> +1

jasnell: propose closing itt
... someone can come up with a more concrete issue

<cwebber2> +1 also

s/itt /it /

<cwebber2> I think it would make sense to have a more concrete proposal

arnaud: can you agree with this direction?

elf: too bad Harry left the room ... because he recently proposed something similar
... also Tantek is not here
... if the Issue as stated is too blurry, it's OK to close

<jasnell> if someone wants to propose using microformats terms in AS2, then they'll need to (a) show examples of how it would work and (b) describing the normative mapping

<jasnell> +1

<Arnaud> PROPOSED: Close ISSUE-15 as is

<Tsyesika> +1


<eprodrom> +1

<rhiaro_> +1

<deiu> +1

<aaronpk> +1

<sandro> +1 but very interested in solving the related underlying interop issues!

<cwebber2> +1

<elf-pavlik> -0 we currently have proposal from harry to include microformat terms in AS vocaba and/or context

<akuckartz> +0.5

<Tsyesika> I think closing it does make sense since it is quite vague but i think the discussion did bring up a good interop issue between AS 2.0 and microformats so it'd be interested in seeing rhiaro_'s produce an issue

<Arnaud> RESOLVED: Close ISSUE-15 as is

<cwebber2> good, will give me a chance to brew coffee and make a quick breakfast :)

<Arnaud> *lunch break for 1h*

arnaud: goal is to write very specific issues, with actions that are clear

<eprodrom> I'm going to be available from 13h -> 14h and then after 16h

<wseltzer> [lunch break]

<jasnell> are we starting again soon?

<cwebber2> sounds like people have returned :)

<Arnaud> yes

<Arnaud> just sitting down

<Arnaud> sorry for the delay, restaurant was a bit slow

<cwebber2> no worries :)

<cwebber2> Arnaud: you all are just trying to savor Paris' gastronimical offerings :)

<Arnaud> :)

<Arnaud> we're still missing a few people

<jasnell> I usually wake up an hour from now

<eprodrom> ha ha

<jasnell> thankfully, my fiancee is taking the kids to school this morning so I can sleep

<elf-pavlik> sandro, re _: in JSON-LD context, just in case you didn't see this closed issue:

<elf-pavlik> scribe: elf-pavlik

Arnaud: let's resume and close as many issues as we can
... jasnell please pick a victim


<trackbot> issue-16 -- better separate grammar/vocabulary and improved spec structure -- open


jasnell: i see it straight forward

bblfish (Henry Story) arrives

jasnell: comes down if we want to split terms into core and extended
... implementers will need to look in two separate places
... i would prefer to keep it together

elf-pavlik: i see grammar / core needing to stay more stable & frozen while extended terms can change and evolve more

eprodrom: i see it fine the way it stays and defer to jasnell

<cwebber2> +1 on deferring to jasnell

sandro: i don't know what i talk about, different namespaces or editorial?

jasnell: i see it as editorial proposal

sandro: fine to defer to editor, if we want to split namespace i would have problem

jasnell: i can separate that if you want


elf-pavlik: grammar terms like as:object, as:actor etc. can drawn in less important terms, but i see more importance in JSON-LD contexts than editorial

eprodrom: i leave it to editor

<eprodrom> +1

eprodrom: i want to see show of hands, how many people are familiar in core vs. extended classes?

<rhiaro_> +1 I think so..

<AnnB> -1

+ i can see diference

<aaronpk> +1

<cwebber2> +1

<melvster1_> +1

<ben_thatmustbeme> +0.5 (read some of it, know what the difference is)

<Tsyesika> +1

<sandro> 0 some but not a lot

<akuckartz> 0.7

<bblfish> hi

jasnell: i will take another look at it and make editorial changes to address this issue

<Arnaud> RESOLVED: Leave it to James to edit as he sees fit and check back with Erik to see if that satisfies ISSUE-16

<eprodrom> +1

<obergix> zhhhhh


jasnell: few weeks ago we resolved that Person object represents profile

<rhiaro_> issue 17 is duplicated in issue 26, is that what jasnell just said?

elf-pavlik: ok with closing it

rhiaro_: duplicated in issue-26


<wseltzer> issue-26?

<trackbot> issue-26 -- Representing profiles in Activity Streams 2.0 -- open


sandro: i will repeat myself
... most of the work in this place i consider wrong!
... people modeling humans who don't want to be model as computers
... existing mainstream models model accoutns

melvster1: facebook models humans

jasnell: PROPOSAL we chose just not to model people and don't define any of actual properties

<bblfish> I disagree with what @sandro said, but it's probably immaterial now

<rhiaro_> +1 on closing 17 -1 on closing 26

jasnell: this will allow us to close issue-26 and issue-17

bblfish: i disagree with sandro, FOAF has Agent and account
... but immaterial to questions

elf-pavlik: +1 rhiaro_

<Arnaud> PROPOSED: Close Issue-17 as superceded by issue-26

<elf-pavlik> +1

<rhiaro_> +1

<jasnell> +1

<deiu> +1

<aaronpk> +1

<Tsyesika> +1

<cwebber2> +1

<akuckartz> +0.8

<eprodrom> +1

<Arnaud> RESOLVED: Close Issue-17 as superceded by issue-26

elf-pavlik: if someone has proposal for multiple profiles please put it out

<bblfish> Issue-26?

<trackbot> Issue-26 -- Representing profiles in Activity Streams 2.0 -- open


<jasnell> I'm unable to hear amy

rhiaro_: if you have profile on one site and want to move it to another site you need way to do that
... i still work on it and can take ACTION to make counter proposal to current Profile draft

jasnell: i don't see need to have our own way to model profiles

i opened issue-26 to decide how to model *Actions* on existing profile

jasnell: we have existing vocabs like vcard

elf-pavlik: also h-card (currently not compatible with AS2.0)

eprodrom: i understand that we have concept of a Person but would like to model that single person has more than one aspect to the world

<rhiaro_> I think in that case I would agree with not going into modelling a profile for AS2, but would prefer to rename it something other than Person, which becomes misleading

eprodrom: my work aspect, my personal aspect, my private aspects etc.

<ben_thatmustbeme> There does not really seem to be any existing system to do this (at least to me), you just create a second account

eprodrom: i see it as interesting way to do social networking but in 2016 single person maps with single acounts

<ben_thatmustbeme> i mean any existing public social network that does this

eprodrom: we could put some support for it as another element ~ aspect, own by person or controlled by person

<rhiaro_> ben_thatmustbeme: and then it's a pain if you want to move things around between networks, right?

<ben_thatmustbeme> I would argue we continue with a single person, but can come back to in later versions

<eprodrom> Are we talking about FOAF?

<eprodrom> I'm confused

bblfish: i know foaf:Person and see it vaguely defined

<eprodrom> Aren't we talking about AS 2.0?

bblfish: something about 4-dimentional time continum (scribe didn't get it ;)
... URI reffering to the person doesn't mean always the same person

<ben_thatmustbeme> rhiaro_, it can be, but I really don't think we should be concerned with adding functionality for that

bblfish: we should keep timeline in mind
... e.g. changing jobs and any other cases

<melvster1_> foaf person can also be an 'imaginary friend'

bblfish: where you want to move identities or have them merged

Arnaud: i have impression that issue-27 relies on resolution of issue-26

jasnell: vocabulary currently has a Person object
... i could say that i added particular piece of information to my profile

<eprodrom> as:Person as:Update as:Person

<ben_thatmustbeme> then again, in indieweb, this is a matter of rel=me to my facebook account. If facebook was able to integrate directly to my site, i wouldn't need that account

Arnaud: do you say that we don't need anything else than we have to describe person?

<Zakim> elf-pavlik, you wanted to ask parklize about sioc:UserAccount and foaf:Person

<rhiaro_> sioc model is largely fine

elf-pavlik: why SIOC defines sioc:UserAccount

<rhiaro_> sioc is missing personas, but don't need to go there now :p

parklize: foaf:Person defines a person and in our perspective one has multiple accounts

<hhalpin> do we have a use-case where this difference makes a difference (i.e. between account and profile)?

in SIOC we use sioc:UserAccount as *author* of a sioc:Post

<hhalpin> My guess is no.

<hhalpin> Until I see one.

<aaronpk> hhalpin++

<Loqi> hhalpin has 5 karma

rhiaro_: fine with not having new model but consider renaming Person to e.g. Profile

elf-pavlik: or Persona

<sandro> +1 rhiaro let's rename Person so people don't keep thinking it's a Person

elf-pavlik: +1 rename Person

<hhalpin> The only use-case I can see is migrating personal profiles from one system to another and connecting them to a single person

<cwebber2> irc client was going crazy, back

<akuckartz> Yes, rename "Person"

eprodrom: the way you think about it in AS2.0 it may seam crazy that actor operates on oneself, but for me it makes a lot of sense

<Arnaud> PROPOSED: Close ISSUE-26, renaming Person as Profile

elf-pavlik: BTW

<eprodrom> as:Person as:Update as:Person

<ben_thatmustbeme> +1 for just "update profile"

<rhiaro_> I think a 'person/profile' updating itself is fine too, the connection to an IRL person isn't necessary

jasnell: issue-27 asks if we need to model specific *granular* updates to profiles

<rhiaro_> I think issue 27 and 26 are merging now

<bblfish> +1 for jasneell

<rhiaro_> in discussion

<bblfish> jasnell++

<Loqi> jasnell has 15 karma

<rhiaro_> elf-pavlik: good point re: profiles for non-persons

<hhalpin> Usually a:Person has a name and age (1983?), not a a:Profile (likely a few years old).

eprodrom: i see you point in having value about say 'this what the delta was - kim removed phone number' but keeping it simple sounds better for now - Kim updated her profile

<hhalpin> It all gets pretty zen pretty quickly "Does a profile have a name?"

<hhalpin> i.e. there is no correct way to do this, it's all context dependent.

eprodrom: i don't like idea of renaming Person to Profile
... how many of us knows the different classes?

<Arnaud> PROPOSED: Close ISSUE-26, use Person for profiles

eprodrom: if we need to represant Organization

elf-pavlik: -1

eprodrom: many add software process as actor

<hhalpin> For example, cases where "profile" vs. "person" make a difference, can we just maybe do this in extensibility and not in the core Vocabulary?

bblfish: we need both Profile and Person

<ben_thatmustbeme> would the ability to have "alternate person that is also me" be possible?

elf-pavlik: you can edit profile *document* which differs from Person

bblfish: let's use FOAF


<ben_thatmustbeme> if a single person in the real world has 2 accounts, they can link to their other "me"

<hhalpin> ^^ Again, we've been down this rathole a few times

<eprodrom> hhalpin: I agree, I think it's some fine parsing

bblfish: we can also use foaf:Person + sioc:UserAccount

<Zakim> elf-pavlik, you wanted to discuss Person's profile, Organization's profile, Event's profile, Venue's profile

<hhalpin> otherewise known as a waste of time IMHO

<bblfish> I

<bblfish> I am for having two terms: Person/Persona and Profile

hhalpin: we can try to design perfect vocabulary

<akuckartz> http-range14++

hhalpin: but we will never agree

<Loqi> http-range14 has 1 karma


<Loqi> akuckartz has 2 karma

<bblfish> the case was made that user story 1 had this

hhalpin: where in approved user stories we need this difference
... we can stay very ambiguous about real name or not
... in favor of not changing anything unless we really need it proved by use case

<hhalpin> I *really* think we will never come to agreement on perfect vocabulary modeling

<hhalpin> that is why we have extensibility

<ben_thatmustbeme> hhalpin++

<Loqi> hhalpin has 6 karma

<hhalpin> so, is there really a difference where we absolutely need a profile versus person difference?

<eprodrom> I have to leave, unfortunately

Arnaud: do we have user story that requires making a difference

<hhalpin> If it's not clear, then let's wait till we implement it and see if it becomes clear

<rhiaro_> implementation++

<Loqi> implementation has 1 karma

<hhalpin> vocabulary definition discussed in abstract will go on indefinitely

jasnell: we don't need to take time to define it precisely

<bblfish> jasnell

... enough that we can say Person updated Profile

<bblfish> one could have just the general notion of a document

<hhalpin> To be honest, I don't really care, w can add it and if no-one uses it by the time we hit CR, we delete it.

<rhiaro_> I agree we don't need both, I just think Person name is misleading

jasnell: just having possibility to mark something as profile makes sense to me

<eprodrom> Thanks, all

<Zakim> elf-pavlik, you wanted to discuss ask aaronpk about semantics of h-card

<sandro> rhiaro, I can live with "Person" as long as we're clear in the documentation that "Person" can be fake, and does NOT refer to a Human Being.

<hhalpin> yep, we need to clear it can be fake

elf-pavlik: can one person can have multiple h-cards

<hhalpin> i.e. multiple personae

aaronpk: h-card doesn't represent an account since has no service associated to it
... it vaguely represents person

<Arnaud> PROPOSED: Close ISSUE-26, use Person for profiles, clarifying that a Person need not be a physical person

elf-pavlik: what about rel="me"

<akuckartz> What if some one want s to model Person in an extension ?

<Loqi> 4azS58uU.jpg

aaronpk: it links URLs i can link to my twitter account which doesn't have h-card

<rhiaro_> +1

bblfish: person Persona as superclass of person

<jasnell> -1 to introducing Persona or renaming Person to Persona

bblfish: we already have documents with ldp:Resource so we can already distinguish people from resources

sandro: fine with keeping Person and cleaning up documentation

<jasnell> +1 on the proposal

<aaronpk> +1

<rhiaro_> +1 with clarifying documentation

<Arnaud> PROPOSED: Close ISSUE-26, use Person for profiles, clarifying that a Person need not be a physical/real person

<akuckartz> -0,01

<rhiaro_> +1


<ben_thatmustbeme> +1

<sandro> +1 with clarifying documentation

<bblfish> -1

<melvster1_> -1

<deiu> +0

<bblfish> remove -1

<hhalpin> "Represents an individual person." -> "Represents an entity that can be communicate, i.e. a person or a personae of a person"

deiu: can we vote on *definition* not on word?

<cwebber2> sure, +1

melvster1_: the way we model thinks in FOAF community, we keep distinction between profile document and person herself

<hhalpin> ^ that's my try of a definition

<rhiaro_> wrt to profiles, a profile can have a different creation date to a person has a birthdate - but if that never ever comes up as a problem, they can be one and the same

<jasnell> fwiw.. I would be +1 to the notion of adding a Profile object type

<hhalpin> Usually when you like a profile, it means you like a person

<hhalpin> i.e. contextual ambiguity is resolved.

melvster1_: we had with tantek this coversation over time
... i may like Rick Martin's page but not him as person

<tantek> sandeep has joined #social


<Loqi> melvster1_ has 1 karma

<hhalpin> Yes, but another person (in fact, most people) would "like" Ricky Martin's page because they like Ricky Martin, not his CSS

deiu: can profile represent multiple people

<ben_thatmustbeme> we are talking range14 but not by name

<ben_thatmustbeme> can we stop

<hhalpin> yep

melvster1: my profile contains me and my public key

<tantek> ???

<bblfish> ++melvster

<rhiaro_> I'd hoped to avoid going this far

Arnaud: do we see this distinction relevant in context of AS2.0 and our user stories?

bblfish: didn't we have user story to like something so melvster1 story makes sense
... we can follow a person but like their personal page not them

<jasnell> fading very fast here. sleep was a long 24 hours ago

<cwebber2> D:

aaronpk: on social web we talk about following people's content not them literaly

<ben_thatmustbeme> Arnaud: we could waste the entire day talking about http-range14 without using the name... but can we do something to stop that

bblfish: in the web you describe people, Cesar different than book about him

<cwebber2> I agree, please move on

<hhalpin> Again, humans resolve contextual ambiguity just fine

<akuckartz> We could resolve this by renaming Person ...

<hhalpin> machines don't.

<aaronpk> i think we need a range14 hammer

<Zakim> elf-pavlik, you wanted to discuss point two different issues discussed at the same time

<ben_thatmustbeme> or at least re-poll and see if we can resolve -1's

<tantek> wseltzer++

<Loqi> wseltzer has 1 karma

<hhalpin> Didn't we have this conversation at our last f2f?

<rhiaro_> elf-pavlik++

<Loqi> elf-pavlik has 21 karma

<AnnB> elf-pavlik++

<Loqi> elf-pavlik has 22 karma

<aaronpk> we need a resolution and to never speak about it again

<rhiaro_> I agree with what wzeltzer is about to say

<hhalpin> No one realistically likes a font

<tantek> hhalpin: indeed - so how about those previous f2f minutes?

<hhalpin> using the "Facebook like button"

<Zakim> wseltzer, you wanted to suggest that we don't need to resolve this. We don't need a way to talk about abstract "Person"

<hhalpin> They've been e-mailed to list, I should update the wiki

<rhiaro_> wseltzer++

<Loqi> wseltzer has 2 karma

<rhiaro_> ++++++++++

wseltzer: we don't need to think as Persona different than Person

<cwebber2> so, it seems like it's possible to solve this either directly or with an extension

<cwebber2> so maybe

<cwebber2> it makes sense to have a document that defines a general process for this?

<jasnell> keep Person. add Profile

<kaepora> What if the profile was for a sentient AI?

<Loqi> 4azJaiQW.jpg

<jasnell> Profile is a description of a Person

<rhiaro_> jasnell++

<Loqi> too much karma!

hhalpin: we can add Profile and if no one uses we just remove it

<Arnaud> PROPOSED: Close ISSUE-26, add Profile

<jasnell> Proposal: Keep Person, add Profile

<kaepora> I think the current policy discriminates against sentient AI :(

<jasnell> +1

<kaepora> Robot rights!

<rhiaro_> +1

<deiu> +1

<melvster1_> +1

<Tsyesika> +1

<tantek> This is why it was a mistake to bikeshed "author" into "actor" or "person" etc.

-0.5 unless Organization, Event and Venue also can have Profile

<cwebber2> kaepora: it doesn't if sentient AI is a person

<cwebber2> +1

<sandro> +1 Okay, assuming we're still clarifying Person IS NOT HUMAN BEING. Otherwise -1

<jasnell> can clarify: Profile is a description of a thing, typically a person

<ben_thatmustbeme> +1 to closing this issue

<akuckartz> +1

<cwebber2> contentious socialwg issues are people too

<bblfish> +1 distinction to a Person and Persona +1 to distinction between those and Documents. But we don't need to define document because we have LDPRs for that

<aaronpk> +1

<AnnB> elf draws diagram

<jasnell> I will write up a description of the Profile object, we can resolve the specific definition of it later. Let's close #26 and move on now

<rhiaro_> elf, stop!

<hhalpin> If someone wants a "real individual human being not a personae" then they can use an extension vocabulary

<tantek> -1 to any distinctions that no one has had to deal with in live public implementations

<bblfish> +1

<hhalpin> I think that's a general purpose observation from tantek.

<sandro> tantek, are you vetoing this proposal, just following on IRC???

<jasnell> lost audio...

<hhalpin> tantek - do you want to kill any addition of a profile?

<tantek> the whole "person(a)/profile" thing is a rathole

<hhalpin> Agreed

<Arnaud> tantek, are you objecting to the proposal?

<AnnB> I realize this is a complicated discussion ... but certain participants bring real-world experience that is not deserving of dismissive tones

<sandro> tantek, you're keeping us stuck in the rathole

<cwebber2> jasnell: I'm here

<AnnB> it

<hhalpin> However, my proposal is we let them keep it, and if no-one can demonstrate running code that uses it by end of CR, then we remove it.

<akuckartz> AnnB++

<tantek> please determine consensus without me

<hhalpin> That would hold, I think, in general for *all* vocabulary items

<tantek> consider my input purely "informative" during the next two days

<AnnB> it's more important to achieve real consensus than to pound people into the ground

<Arnaud> forget: Close ISSUE-26, add Profile, clarifying that a Person need not be a physical/real person

<bblfish> +1

sandro: Arnaud used resolution not matching proposal!

<Arnaud> RESOLVED: Close ISSUE-26, add Profile, clarifying that a Person need not be a physical/real person

sandro: i take it back

<ben_thatmustbeme> +1 to clarication


<trackbot> issue-27 -- How do we represent changes to a profile in an activity -- open


arnaud: does adding profile solves this problem?

jasnell: yes it solves it

<Arnaud> RESOLVED: Close ISSUE-27, revolved with addition of Profile

<AnnB> Claudio Venezia joins (Telecom Italia)

<hhalpin> I think most people can't handle httpRange-14, especially late at night

jasnell: i prefer Evan to talk about issue-31

<ben_thatmustbeme> issue-33

<trackbot> issue-33 -- Deprecate or remove the "rating" property -- open


<bblfish> Issue-33

<trackbot> Issue-33 -- Deprecate or remove the "rating" property -- open


<cwebber2> +1

<cwebber2> to deprecate / remove

<elf-pavlik> +1

<rhiaro_> microformats has rating, but defined differently

<rhiaro_> extension seems like a good idea

<ben_thatmustbeme> +1 to move to an deprecate / move to extension

<rhiaro_> then people can define however they like

<cwebber2> I don't think rating is used in pump, but like is

<cwebber2> like is used

<cwebber2> there's no rating used afaik

<ben_thatmustbeme> cwebber2: i doubt we are going to remove like, just rating

<cwebber2> Tsyesika, is that right?

<Tsyesika> pump doesn't use it

<cwebber2> Arnaud: jasnell: I think Tsyesika should be able to answer this also

<Tsyesika> but also if we wantd to in the future we can always use extensions

<cwebber2> and she just did

<cwebber2> deprecate it, yeahhhhh

<cwebber2> +1

<cwebber2> again :)

<Tsyesika> +1

<rhiaro_> +1

<ben_thatmustbeme> +1

<tantek> when in doubt leave it out

<Arnaud> RESOLVED: Close ISSUE-33, deprecate rating


<tantek> (why not remove?)

<Loqi> tantek has 186 karma

<ben_thatmustbeme> issue-34

<trackbot> issue-34 -- Context vs. Motivation -- open


jasnell: currently vocabulary has property *context*, intentionally vague
... intended for you to describe context in which activity took place

<cwebber2> additionally, "context" is confusing because of name collision with json-ld

<elf-pavlik> +1 context dangerous to use cause of colisions

<aaronpk> i was wondering about that

<tantek> (kind of like ID did back when?)

jasnell: context has a lot of meaning on itself

<tantek> (is context meaningful without context?)

jasnell: in web annotation model an annotation has motivation

<cwebber2> tantek: we'll define the @context in the context for full context

<cwebber2> oh sorry

<cwebber2> context in the @context for context

<cwebber2> I confused myself.

<cwebber2> ;)

<tantek> cwebber2++ for unintentionally making the point ;)

<Loqi> cwebber2 has 28 karma

<cwebber2> :)

jasnell: too late for me to describe it clearly
... we need to adapt to annotations model in AS2.0

<cwebber2> I think, at least, it's a good idea to drop "context", but then possibly consider whether to incorporate motivation

<cwebber2> but yes

<cwebber2> I agree

<hhalpin> I'd like to see how Web Annotation uses it - I'd kick that decision to the Web Annotations WG

<cwebber2> jasnell should be able to be awake :)

<tantek> has anyone collected the annotations use cases? we keep requesting it.

jasnell: let's push it to later when i can concentrate more

<cwebber2> tantek: not sure, but they commented on the activitypump tracker

Arnaud: i don't see reactions and we may need to accept that's it for today

<tantek> postpone anything to do with "annotations model" until we have permalinks to annotations use cases

<Tsyesika> cwebber2: you mean asleep right?

jasnell: i appreciate all the inputs

<hhalpin> +1 jasnell

<Tsyesika> jasnell++

<hhalpin> Great job staying up!!

Arnaud: thanks you for staying all night

<Loqi> jasnell has 16 karma

<cwebber2> thank you jasnell !

<elf-pavlik> jasnell++

<tantek> use cases first, data model second


<Loqi> jasnell has 17 karma

Arnaud: that's it for AS2.0 for today

<bblfish> jasnell++

<Loqi> jasnell has 18 karma

<cwebber2> Tsyesika: well I meant that jasnell should be able to be awake when we address it, but yes that means he should sleep now :)

<cwebber2> jasnell++ indeed

<Loqi> jasnell has 19 karma

Arnaud: agenda calls to talk about social API and tomorrow day ends earlier

<Tsyesika> cwebber2: oh okay

<ben_thatmustbeme> phone call is losing audio

<ben_thatmustbeme> doesn't sound like anything important anyway

<cwebber2> I would prefer to do the specification walkthrough tomorrow when evan is around right?

Arnaud: let's start discussion on protocol


<aaronpk> i still can't see that page

Arnaud: we lost Evan, can we discuss protocol without him?

<ben_thatmustbeme> i can't see that page either

<ben_thatmustbeme> not sure we are supposed to aaronpk

Arnaud: we have 3 main proposals, how do we get to resolution on how do we go
... we saw some demos already last time

<cwebber2> +1 deep dive into proposals

Arnaud: we need to deep dive into different proposals

<cwebber2> but, it would be nice to maybe do activitypump tomorrow, at least

<aaronpk> ben_thatmustbeme: maybe, but i got added as a guest for the event so I assumed i'd be able to see the schedule

Arnaud: who feels ready?

<cwebber2> so that evan can be around

<rhiaro_> aaronpk++

aaronpk: i have documents explaining how to implement 4 user stories using indie web stack

<cwebber2> though I think if needed, Tsyesika is totally capable :)

<cwebber2> ooh, thanks aaronpk

<cwebber2> aaronpk++

<cwebber2> :D

Arnaud: let's start with it now and others can take it as inspiration

<cwebber2> aaronpk: is there a link I should pull up locally?

<Tsyesika> cwebber2: aaronpk is way mroe propared it seems? lets let him go first

<cwebber2> or just look at the spec on the indieweb wiki

<cwebber2> yes

<Tsyesika> :D

<cwebber2> I agree, aaronpk going first is good


<cwebber2> is great btw aaronpk

<aaronpk> that's the one


<Loqi> aaronpk has 792 karma

elf-pavlik: i will not scribe what document already explains, just reactions of others

aaronpk: displays

AnnB: maybe Tsyesika could also prepare equivalent example for ActivityPump

aaronpk: at least User posts a note

melvster1: we have two concepts - direct identifiers and indirect identifiers

<AnnB> not to bring up the "person" discussion ... I'm interested in whether or not that will play out in the Internet of Things

<elf-pavlik> TimBL recommended avoiding use of indirect identifiers

hhalpin: acct: or mailto: identifiers identify accounts <- indirect identifiers

melvster1: facebook uses in graph protocol # identifiers

<AnnB> if "person" is not necessarily a real person ... then do we need to consider this definition in light of the vast IoT complexity coming down the pike?

<AnnB> (people seemed to say yes, we do)

deiu: situation looks differently when you search centralized repository

aaronpk: it changed from 2003 where people identified themselves by URL, now they even identify themselves by #hashtags
... i'll illustrate some of the user stories with MicroPub, Webmention, Microformats and sometimes PubSubHubBub
... User posts a note as very basic and fundamental

<AnnB> protocol walk-through ....

aaronpk: Replies as very complex and representative

<hhalpin> Note that the Graph API is developer-facing, not user-facing. Anyways, we should just support *any* URI scheme for identifer.

aaronpk: SWAT0 as the only one officially approved

<hhalpin> Although there is not yet a URI scheme for hashtags.

aaronpk: i also leave out authentication and assume that you have *Bearer* token

sandro: what about new line characters?

aaronpk: i need to add explaination
... you can also return short url in header, also some people implemented slug as parameter


... mp- scopes parameters as not of object that gets created

sandro: the reason you don't use PUT?

aaronpk: in general we always use GET and POST
... more like RPC system
... you interact with the endpoint not URL of resource you try to modify
... it offloads work to separate system, in this case micropub endpoint
... which one can use somewhere on separate server

bblfish: you could gave different systems which fit with your system
... you could link also to RESTful endpoint via link relation similar as you do with rel="micropub"

aaronpk: at this point RESTful doesn't seem like a goal, especially that nowadays what people call REST actually doesn't do REST

Arnaud: what do i get when i do GET on this URL?

aaronpk: HTML with microformats
... but you can look at parsed version

Arnaud: so your endpoint added a lot of metadata to the original entry

aaronpk: yes generated object can be quite different e.g. include published date
... other story has minimal HTML example

Arnaud: this parsed version has microformat data extracted from HTML
... does endpoint has full freedom in what to add / remove to the original data?

<ben_thatmustbeme> Should note that some others (me) use get with url= to get the source (form encoded) object

aaronpk: you can get source of the post you intend to edit
... rhiaro has example with using markdown and getting markdown back from endpoint
... we try two different ways for that currently
... if for some reason HTML didn't include content, then retrieved source still should have it

Arnaud: how does followers know that you posted something?

aaronpk: Responses example will explain it
... we support plain forms since it makes it easy to debug and see what happens

<rhiaro_> jsdr++

<Loqi> jsdr has 1 karma

aaronpk: all programming languages support it, even embedded devices

bblfish: i wrote a proposal to do pingback in this way
... people really didn't like javascript at that time
... i see you trying robots interact as well as user friendly, kind of low level browsers
... we don't want to write specs for particular vocabulary

elf-pavlik: related issue

bblfish: how do you extend it?

aaronpk: examples aim to demonstrate user stories as precisely as i could read them
... document exists explaining how it works in detail


aaronpk: micropub doesn't require to change anything when you want to add new terms to vocabulary
... for the basic case, every microformant endpoint supports h-entry content
... my endpoint has custom support to posting food types of posts
... but if someone uses my app other people's endpoint will not undersand it

<cwebber2> but there's no way to specify the vocab?

aaronpk: challange is, if someone doesn't support my extension - how the app should behave ?
... should they render everything as text etc.

<cwebber2> what happens when you have a server that accepts "runs", but it thinks it's about a program running

<cwebber2> and your server meant to track someone running a mile?

<sandro> sandro: capability discovery is one more option

<ben_thatmustbeme> Cwebber. If it's not part of the standard it should be app prefixed

elf-pavlik: does challenge come with side effects?

aaronpk: no, only that you don't know how to display it

<cwebber2> ben_thatmustbeme: how does app prefixing work?

<ben_thatmustbeme> X-myapp-run

<ben_thatmustbeme> As the field name

<cwebber2> ben_thatmustbeme: hm

<ben_thatmustbeme> I believe aaronpk uses x-p3k-food for example

sandro: that argues to post microformats, so server can just display them

aaronpk: if you understand the h-entry vocabulary then can render it
... i would like to get to the long one

Arnaud: more general conversation we can keep for later

<ben_thatmustbeme> Sandro, we have had a few discussions of being able to discover support. I'm in favour of listing all types supported on a query for it


aaronpk: cropping happens outside of those tools
... design based on microformats vocabulary where photo gets put in *photo* property

<AnnB> aaronpk: this is essentially an example of how to do a post with non-text content in Micropub

aaronpk: disclaimer in practice no one write that multipart noise by hand, libraries do that!

sandro: we can attach simple html5 app showing how to do that or at least curl commands

bblfish: why do we need to have it standarized?

<sandro> sandro: that is, a snippet of JS, hopefully 1-5 lines.

aaronpk: interesting point and it relates to capabilities discovery

bblfish: do you want robots to read forms, while nowadays people do it

Arnaud: h-entry means "I want to make a post" ?

aaronpk: if you don't add publish date it will use current date
... if you want to name a photo you just add another parameter

<cwebber2> user stories seem to be often that way :)

aaronpk: i've learned that those user stories stay vague and we need to clarify them

<cwebber2> I'm glad we have them!

elf-pavlik: ISSUES please in

<cwebber2> is vagueness a feature? discuss! ;)

aaronpk: it doesn't matter where endpoint stores the photo! it can upload it to S3 bucket whatever

Arnaud: so it creates two resources 1) with binary 2) with information about photo

elf-pavlik: please see

aaronpk: endpoints can do whatever they like with the photo
... response doesn't give you URL to encoding .jpg but you can GET the resouce and find it in HTML+Microformats

Arnaud: it touches point PUT versus POST

aaronpk: you can't put to photo URL since it can have different credentials which only endpoint can use


elf-pavlik: we can think of h= as type=

aaronpk: you could also use mp-h=
... h = upside down micro symbol - Microformats
... mp-h= could make more sense since you don't use it in your content

<cwebber2> aaronpk: this is really helpful btw

<cwebber2> aaronpk: there are some user stories not brought up here, which I think might be interesting to address



aaronpk: SWAT0




... i changed names to use names of people participating in IndieWeb community

<cwebber2> those are all from "entirely positive"

<rhiaro_> in microformats category = tag

<cwebber2> ben_thatmustbeme: ok, thanks

<cwebber2> oops

aaronpk: we use the same property *category* as we would use for text tags

<cwebber2> auuugh

<cwebber2> ben_thatmustbeme, thanks ;p

aaronpk: i wish we would name it just *tag*

<cwebber2> ben_thatmustbeme, unfortunately that's what my irc client does by default

aaronpk: but in this case you use person URL as value instead of string

sandro: does people assume that @aaronpk means twitter username?

aaronpk: in a way, it got syndicated to twitter and than mentioned person correctly

Arnaud: where did the endpoint got "People" information from?


<ben_thatmustbeme> My test of tagging tantek

elf-pavlik: you could have more specific property which would imply value a person

aaronpk: please consider this as early exploration with person tagging

<cwebber2> It's not immediate

<cwebber2> but

<cwebber2> I put it on IRC above:

<sandro> sandro: maybe think of it as link-based tagging (cf keyword tagging) , rather than PERSON tagging

cwebber2: first of all THANK YOU A LOT aaronpk - very helpful!

<ben_thatmustbeme> aaronpk++

cwebber2: i know you didn't prepare but can you comment on those stories?

<Loqi> aaronpk has 793 karma

Arnaud: let's wait for aaronpk to finish with this story

<cwebber2> sorry!

<ben_thatmustbeme> cwebber2, give me 20 minutes to get to the office and I may be able to

<cwebber2> ben_thatmustbeme++ thank you :)

<Loqi> ben_thatmustbeme has 69 karma

<ben_thatmustbeme> Several of them are not worked out yet. But I have some ideas for sure

aaronpk: we can't assume that Barnaby goes to Bent's site to see the photo

<tantek> what is mp-?

aaronpk: this story didn't have act of starting following but other story has it

... micropub non content property AFAIU

<tantek> (oh right - thanks elf-pavlik )

aaronpk: step 5 - most attempts in SWAT0 failed here
... the saddle thing ... i need to loo at
... tantek didn't subscribed to ben but he still received notification

<sandro> sandro: maybe split SWAT0 final point into to -- the tagged person and the subscibing person

hhalpin: people tried to do that with Salmon but possibly only Evan and one more person ever implemented it

<hhalpin> notes the prior art is here:

aaronpk: mentions difference between tagging on the photo or mention in the post


<hhalpin> It doesn't do the summarizes, but. it worries about distributing and re-distributing comments, ordering, etc

obergix: you can't comment on the photo and you could get ddosed

<AnnB> I appreciate aaronpk's articulate aand calm presentation and answering of questions

<cwebber2> AnnB: I agree!

aaronpk: Responses gave me really good exercise, i encourage everyone to try implementing it!


<cwebber2> aaronpk is doing great

aaronpk: I find Inbox stream loaded term and sounding specific
... explains PubSubHubBub
... need to write not explaining how we know that we deal with like

<elf-pavilik> rhiaro++

<rhiaro_> elf-pavlik I already did it

<Loqi> rhiaro has 71 karma

<ben_thatmustbeme> sandro, "<Zakim> -INRIA" i'm gone

<cwebber2> I don't hear any audio

<Arnaud> cwebber, we need your help reconnecting us

<ben_thatmustbeme> cwebber2, the conference just closed, sandro is working to fix it

<cwebber2> Arnaud: how can I help?

<sandro> Pi code Organisateur: 21830239, Pin code Participant: 23695487)

<kaepora> Hey everyone, what's the plan until dinner?

<kaepora> (Sorry, I'm at my office downstairs)

<ben_thatmustbeme> oh... i'm getting some error audio

<sandro> ben_thatmustbeme, type 23695487#

<Arnaud> ben?

<Arnaud> we need you to help by entering the code

<ben_thatmustbeme> i tried

<Arnaud> ok, let's try again

<Arnaud> when you hear the french prompt, dial in the code

<sandro> ben_thatmustbeme, type 23695487# again ?

<cwebber2> Oh I see

<ben_thatmustbeme> yes

<cwebber2> I can hear

aaronpk: we talk about notifying others about like

<ben_thatmustbeme> cwebber2, is currently done in a reader. I want to move the list of who to follow in to users own sites (a following list) then use micropub to create an entry with h=card&mp-action=follow&url=<feed url>

<ben_thatmustbeme> cwebber2, does not need micropub, thats just polling the person's feed url (parse microformats if its in some sort of reader perhaps)

<rhiaro_> ben_thatmustbeme, cwebber: I experimented with follow posts at

<rhiaro_> with a u-follow-of property or something

<cwebber2> ben_thatmustbeme: rhiaro_: thanks!

<ben_thatmustbeme> cwebber2, aaronpk has already been able to do this, when he posts the note it does he sets private entity, i don't know if he does that through micropub, i dont' think so. but could be, and would be cool. if he shares a note with me, it needs to be given an auth token when polling / viewing to know its me, then its just a matter of notifying (webmention

<ben_thatmustbeme> but when it polls it will get Denied, so it tries again with auth token, now it knows its private)

<cwebber2> rhiaro_: I do like that you have sparql on pages :)

<rhiaro_> cwebber2, ben_thatmustbeme: so I do a regular micropub post with the experimental follow-of property, not an action

<scribe> ACTION: pelf to draw Follow vs. Subscribe with account having multiple feeds allowing subscription independently [recorded in [[1]|]]]

<trackbot> Created ACTION-60 - Draw follow vs. subscribe with account having multiple feeds allowing subscription independently [on Pavlik elf - due 2015-05-11].

<ben_thatmustbeme> cwebber2, on that one check out

<cwebber2> ben_thatmustbeme: thanks, looking

<ben_thatmustbeme> cwebber2, finally on this is just a reader really (push consumer)

aaronpk: we don't follow REST here...

<ben_thatmustbeme> some parts of that one not worked out there

aaronpk: twitter just implemented "quoted" tweet
... facebook has share button and let's you add extra content

<cwebber2> ben_thatmustbeme: as being done by just reading an rss/atom feed also doesn't work with private communication though

elf-pavlik: could you use share-of property?

aaronpk: we need to gather more experience for that

<cwebber2> it does assume that all communication is public, right?

elf-pavlik: facebook show shares of post

obergix: if i share privately it works differently

aaronpk: no stories include private content so far

sandro: i would like my site to make a copy of it

<cwebber2> I suppose it might be possible for the inbox to have a feed that's fetched separately with a bearer token though

<ben_thatmustbeme> cwebber, think of it more as private messaging being direct messaging in twitter or message in FB. they never appear in streams

<ben_thatmustbeme> but you can always pass an auth key and the persons feed could include any posts you are allowed to see

<obergix> applauses

<cwebber2> aaronpk++

<Loqi> aaronpk has 794 karma

aaronpk: stories need some clarifications before freezing them

... please use:

<cwebber2> ben_thatmustbeme, I could probably configure ERC to do it

<cwebber2> ben_thatmustbeme, I might even be able to script it to behave differently on this network, but I find that idea to be annoying :)

deiu: ^ we can go from top

<cwebber2> so, 5 minute break was it?

<cwebber2> I can't tell from over here

Arnaud: we can write same as aaronpk did so we can compare them side by side

<ben_thatmustbeme> aaronpk, cwebber2 and i ask how long of a break is this?

<aaronpk> nobody is really sure, everyone wanted a break

<aaronpk> probably like 10 minutes

<cwebber2> so, is the meeting still happening at this point?

<cwebber2> or is it just socializing mostly?

<aaronpk> just chatter right now

<cwebber2> ok! just wondering if I'm staying around or not

<eprodrom> Are we on break?

<cwebber2> eprodrom: yes, though I don't know if the break is being resumed from

<eprodrom> Gotcha

<eprodrom> So what'd I miss?

<eprodrom> Are we doing all Turtle now?

<cwebber2> eprodrom: 100% turtle delivered over SOAP

<ben_thatmustbeme> eprodrom, have a quick question for you for now

<eprodrom> OK

<eprodrom> cwebber2: great

<eprodrom> ben_thatmustbeme: shoot

<eprodrom> (metaphorically)

<ben_thatmustbeme> pulling up.. one sec, it was a clarification on a user story you created

<eprodrom> Oh, cool.

<cwebber2> aaronpk did a really clear job of demos tying the protocol to user stories

<aaronpk> hey eprodrom! yes i had some questions


<cwebber2> that's what you missed :)

<cwebber2> whoo resuming

<ben_thatmustbeme> on part 2, "and others" meaning Ben's followers or alexa's followers?

<rhiaro_> scribenick: rhiaro_

Arnaud: harry has been on the queue for a while

harry: Would like to structure next discussion
... It's obvious micropub covers the first 4 user stories really well, impressive
... We shouldn't question too much technical choices in user stories
... tomorrow, don't drill too much down into technology

<eprodrom> ben_thatmustbeme: yeah, good question

<eprodrom> At least Ben's followers

harry: but there is a time and a place to talk about differences in methodologies
... now would be a good time to do it

<eprodrom> Probably also the audience for Alexa's video

harry: q&a about mircropub design issues
... some comments:
... 1) why was federation an optional item in charter?
... there was some PuSH/Salmon stuff in the background
... opening up spam problems, privacy and security concerns
... coudl be really hard to deliver
... that's why it's a separate deliverable
... was interesting that micropub ends up using PuSH

<ben_thatmustbeme> eprodrom, aaronpk and I were going back and forth on how to model that, i figured ben's followers, which meant a "like" post, but the other can be done from PuSH

harry: useful to discuss if that's something we want to keep
... most crypto/security stuff I'd like to keep in federation deliverable

<eprodrom> ben_thatmustbeme: It's a tricky problem

harry: We're here at INRIA with the security group
... Interesting to see not REST in micropub

<aaronpk> eprodrom: the question was more what was your intent in writing the story

harry: Issues around how to fit in with everything else, seems microformats-centric
... this mp- mechanism. Coudl we have an equivalent url based thing going on
... for extensibility
... Also mf2 does have a single algorithm mapping to json and back

<eprodrom> aaronpk: yes, and like I said it would definitely be Ben's followers and probably the audience for Alexa's video

harry: in case that helps bridge the json-ld gap to microformats


harry: everything you did doesn't use json

<eprodrom> aaronpk: the second is much harder

harry: the demos were not doen using json, but could be doen using json
... the extensibility vocabulary question too

aaronpk: the extensibility thing, it's been addressed
... mf2 defines a parsing scheme to turn html into a data structure. parsing scheme does not care about vocabulary
... could parse html into a totally fine datastrcutre that is meaningless to a consumer

sandro: is it psosible to turn all json into microformats algorithmically?

aaronpk: no

sandro: things that look at the basic microformats structure can turn that into html?

aaronpk: pretty sure that would work
... should be able to recreate an html equivalent, might not be the same as original
... So for extensibility, that's in the vocabulary
... by all means, create new kinds of h-entry posts and add arbitrary properties, and parsers will work
... but consumers might not be aware of new ones
... Extensibilitiy in terms of vocabulary works with that
... With micropub, I wanted to do the same thing
... The syntax portion of micropub is the same as syntax of microformats, so it can be extended withotu changing parser level of micropub
... that's the reason for mp-*

... is to scope all control-level params into their own bucket, so everything else is considered content

... if you wanted ot add new property to h-entry, like foo, you can do that in your html
... it'll show up in parsed version
... if you want to create using micropub, you can create new h=entry with foo=bar property, server will understand
... generic micropub endpoint could take anything you put in and parse out in html
... would be bidirectionally compatible

elf-pavlik: why mp-* not headers?

aaronpk: headers usually harder for clients
... in a form you'd end up with hidden form elements
... for control things
... like tokens and junk like that
... parsing headers is usually more work, and differs depending on server

harry: re: extensibility I agree in theory that works but needs workign through

sandro: this is json level extensibility not json-ld
... json-ld means two extesnions are never going to accidentally do the same property

bblfish: I can see harry's point
... Really happy to see this stuff
... even though we might have to put into different groups of standards
... important for different groups to show complete story that includes federation
... everyone should do this
... distinction about federation isn't clear
... these demos of stories help

Arnaud: intially the draft was the way it was for fear of being a target we could never reach
... but if we sovle both problems at once, nobody is going to complain

harry: nicest thing about micropub is that it's very simple

<elf-pavlik> eprodrom go ahead :)

<cwebber2> you go ahead! no you go ahead! :)

<aaronpk> we're in france, not portland

eprodrom: there is clearly a great use case for having a client api regardless of federation
... doesnt' need to be federated from the get-go
... value in having a social client api
... we documented half a dozen or more social apis shows that a lot of peopel are doing this
... second thing: are we going to standardise this payload format which is almost entirely different from the payload format that we have as our social data syntax
... what good reason for this? I'm not convinced that json based apis are too hard for developers
... there are a ton

<Zakim> elf-pavlik, you wanted to discuss responsibilities of servers and clients (yes plural e.g. content editors and readers/browsers)

elf-pavlik: would like that we all try to very clearly describe which responsibility goes where
... in different architectures, eg. micropub assumes server can render data
... ldp doesn't make this assumption
... Made an issue about specifying assumptions
... Which responsibility where?
... Same with side effects
... Server, or client?
... And specify which approach puts them where
... For example, I dont need my server to render something
... not just one client one backend, we could have reader, editor client, different backend, different servers

<cwebber2> eprodrom, fwiw, I agree that there's value in having a distinction between client->server stuff and federation, though I don't agree that means having to deliver two different documents

elf-pavlik: for indieweb I can have webmention on one server, micropub on different domain
... with PuSH there's a third

<cwebber2> eg, mediagoblin has implemented the client->server, but not the server<->server stuff

elf-pavlik: So be very clear which componants have which responsibility

<cwebber2> but I think there's a lot of value in having us consider the whole thing altogether

elf-pavlik: And with content type, indieweb reads everything as html

<cwebber2> even though we only got have of it done

elf-pavlik: there's nothing for if the client requests json

<cwebber2> so far :)

elf-pavlik: push updates also html?

aaronpk: PuSH works with json if you want

elf-pavlik: json has advantage of being one format that goes in both directions

bblfish: in the last meeting we agreed we're not here to create one standard yet
... trying to see how these different systems work, learning from each other
... what we have to remember is we're going to have to be able to communicate across these systems, we already have so many standard syntaxes
... now we have a microsyntax for queries
... impossible for anyone trying to build a system to learn all these different things
... for someone who wants to be able to pull in a library to work with different systems, would be some kind of semantic description of these properties
... my system process json-ld, when it reads form encoding, it could know this means this
... a translation from attribute-value encoding into rdf graphs
... if that existed, you could make it massively extensible
... or if you want to post, you can find out how to fit pieces of information into some slots
... you know that the result of sending these attribute values will send to your graph

aaronpk: documentation?

bblfish: machine readable documentation

harry: i would call it a shim

bblfish: has to be automatic

AnnB: that seems hugely difficult

bblfish: I can try to think about that

sandro: json-ld context for microformats vocabulary
... ?

bblfish: for the attribute value query thing

aaronpk: it's the same vocabulary as microformats

<AnnB> I'm not a developer .. so you guys should say if that's hard or not

sandro: json-ld context for that would give you your semantics

bblfish: woudl be nice to have a generic way to extend vocabulary

<cwebber2> in other words, rhiaro should keep working on her work for mapping microformats<->rdf :)

sandro: microformats believes in unified central vocab

bblfish: we woudl like to be able to communicate
... they don't have to maintain it
... could think of attribute-values as sparql queries
... {upside down sparql}

sandro: shim is code that translate from one data format to another

elf-pavlik: can you make a developer user story?

<AnnB> on airplanes we try to reduce the numbers of shims

<eprodrom> !!!

sandro: james said extensibility is essential

<eprodrom> AnnB: in software, too

sandro: I think he meant decentralised
... people can extend without permission
... still with interoperability

<AnnB> extensibility seems important to me

<AnnB> and ditto interoperability

sandro: still not collide with other people accidentally

<eprodrom> That is a great way to put it

<AnnB> but maybe that's nirvana and not realistic

Arnaud: from IBM, our interest is not developing facebook
... Connections not our primary target
... we have products in many different areas, more to do with workflow management
... completely different architecture

<AnnB> +1 from my company

Arnaud: not going to be rendering html the way you would do
... that's why extensibility is very important
... different types of activities

sandro: eg. travel requests, multiple vendors
... so you could end up with a collision

<eprodrom> Isn't that why we have standards groups?

Arnaud: if it's by domian you could avoid collision

<ben_thatmustbeme> "still not collide with other people accidentally" can be difficult to say for certain depending on how the extensabiilty is done, but its good to get it as rare as possible, x-p3k-food will be very unlikely for someone else to use p3k prefix on things

<eprodrom> That's what namespaces are for

aaronpk: if you're using say two different travel request systems that are correctly namespaced, they dont' work with each other?

sandro: correct

<ben_thatmustbeme> yes, but then namespaces have to be something centrally controlled, (url) etc

aaronpk: namespaceing makes it less likely for interoperable implementations to exist
... it's easier to deviate and never merge

sandro: still economic and social pressure to merge

<eprodrom> ben_thatmustbeme: Yes, URLs are "centralized", but that's pretty reasonable to deal with

<eprodrom> DNS is about as distributed as it's possible to do

aaronpk: html, with browsers implementing same tag names that all behave slightly differently

sandro: some of us think decentralised extensibility is absolutely essential as background

<ben_thatmustbeme> p3k is a namespace, but its just defined by the user, without need to any sort of namespace specifically defined

sandro: I'm trying to let go of that and put it on the table

harry: in the charter we have uri based extensibiliity

<ben_thatmustbeme> it was done to keep simple, like css prefixing

harry: we're using json-ld and as2
... aaron said we can fit it into micropub, what else is missing?

aaronpk: the only way to get non-conflicting extensibility is to use prefixes (on strings)
... no definition at the top, not formal

sandro: so it's weak

Arnaud: queue time

harry: chartered deliverable. every system in order to communicate will have some level of shim

<AnnB> aaronpk: example: in extending new things for my food postings, I prefix them with p3k- to make sure to distinquish those from anyone else's extensions

harry: that being said we need convergance on the api
... Just a reminder.

eprodrom: aaron, if the group decided to vote today that micropub woudl henceforth use activitystreams json or turtle as payload, that wouldn't change micropub, right?
... you'd contineu using whatever you feel like using?

<sandro> sandro: we could do or or

eprodrom: Our decision right now is not to take micropub or leave it

aaronpk: not absolutely
... definitely some things I would not want to deal with, but open to suggestions

<elf-pavlik> coupling with microformats issue -

eprodrom: potentially for us two tracks, where we recommend micropub but we also build json based social api more inline with charter
... we'll use the social data syntax in our social api
... that's one possibility
... Building a bridge seems like a failure to come up with a standard
... What I/m proposing is, if we're jammed up on payloads, extensibility etc, then we concentrate on providing an api for social interactions
... and see what happens in the future

Arnaud: typically as a wg we would define standard and if there are bridges that need to be developed that would not be done by the wg
... if people are interested in transitioning to the standard we would build bridges if there was value in that

<Zakim> elf-pavlik, you wanted to shortly mention not obvious *advantage* of using URIs for vocab terms which comes of advantage in decentralized deoployments


elf-pavlik: check social syntax page
... where I try to distinguish between mf vocab, seralisations and data model
... we could have microformat vocab serialsed to json
... we could possibly also use microformats in rdfa (doesn't make sense, but)
... it's not one brittle thing, could extract parts

<sandro> +1 microformats vocabulary serialized in RDFa! (j/k)

elf-pavlik: Also advantage of using URIs for vocab terms that might not be obvious
... if you just use strings for uris you can't dereference them
... if you make small updates to vocabulary you need orchestrated migration to deployments
... advantages to being able to dereference vocabulary uris
... you can communicate with a term
... we can look at advantages and disadvantages
... this is something microformats doesn't provide
... we cannot always say 'developers dont do this so it doesnt make sense'
... most developers work with centralised systems
... should consider a few nuances

Arnaud: good point
... not all or nothing type of choice
... Architecture, which agents are involved, where you make your post, how you make your post, what format
... all this can be subject to different discussions

harry: just reminding people to go back to charter
... I love the three or four ways we can embed rdf into webpages (it's a standardisation failure)
... mf is the larges empirically
... then questionably rdfa
... (according to last neutral academic study)

<ben_thatmustbeme> google CREATED rdfa, so thats something of a questionable source

<ben_thatmustbeme> no, wait, that was schema

harry: Arguement should be do not have embedding data in the charter. Charter is json/json-ld. Should probably stick to moving json around

<Arnaud> :)

harry: But processing into html is possible
... you could skip straight to that
... but if you recieved json, you could read that into html

<cwebber2> ben_thatmustbeme: yeah, RDFa had a lot of work from non-google entities

<cwebber2> including creative commons

harry: serialisation are not blockers, they're in the charter
... extensibility is uri based in charter
... would be good to hook in context to uri in micropub
... More discussion needed about REST
... pretty big difference

<cwebber2> comments about being mostly google-oriented, that's true I think :)

harry: And how we name endp;oints

<ben_thatmustbeme> rdfa i have seen a lot in academic places, library sciences seemed to use it a lot, as well as other standards too

harry: More convergance between activitypump and micropub

aaronpk: no they can be whatever you want, discoverable

sandro: that's interesting
... pretty straightfoward bridge
... to say that ..

<Arnaud> has endorsed rdfa along with microdata

sandro: in LDP we use PATCH for modifying in place without changing what you don't need to change
... PATCH is new http verb
... before it was widely supported you did PATCH over POST
... so I imagine you have an endpoint where you can do all yoru things via POST or you can do restful to other uris
... if you want to support restufl operations you do, if you don't you say here is endpoint for all restful operations
... I think that would let people do both

aaronpk: harry I think I see what you're getting at
... I do want to point out that there are different pieces of how to do things
... Certain aspects of those that I am more and less willing to give up

<ben_thatmustbeme> as far a "social" data though, i know twitter has started to add microformats in to in user profile queries. apparently thats thanks to Known requesting it

harry: overview preferences?

aaronpk: i can try
... in context of looking at something else like activitypump

Arnaud: is there something you'd rather choose to do differently? We should wait to see the rest

aaronpk: I'd like to see the others closer
... comparable situations
... Maybe one of them figured out something that's way less awkward with one of the concerns I was working with
... What I did was constraint based

sandro: static site constraint explains half the differences

<Zakim> aaronpk, you wanted to discuss the "all-or-nothing" re: micropub

Arnaud: plan to close in ten minutes
... burning desire to say more anyone?
... Agree on what we're going to do tomorrow
... Plan is to start around the same time#
... Gathering at 090-0
... trying to start at 0930

<Loqi> I added a countdown for 5/4 9:30am (#5680)

<cwebber2> I'll probably be able to get on about the same time I did today

<eprodrom> No

<eprodrom> Tsyesika will

Arnaud: Assume evan will take us through activitypump

Tsyesika: I will

<eprodrom> Sorry, thought that was clear

Arnaud: up to you who starts

<eprodrom> happy to contribute though!

Arnaud: LDP or ActivityPump

<cwebber2> I would prefer if LDP went first

Arnaud: Idea is to go through two others, then time for demos
... Try to dive into every proposal, step back and try to compare them based on better understandings of appraoches
... Hopefully we don't kill each other, and have enlightenment
... Ideally we will have a sense of where we might go next at the end of the meeting

<cwebber2> but of course those presenting's preferences are higher :)

sandro: aaronpk, groups in micropub?


<sandro> Jake sees in his social stream a video from Damon. Damon and Jake are both in the "Boxing Fans" group. Damon posted the video to the group.

aaronpk: I do have answer
... Using aggregators as groups, I can make an example

bblfish: If we look at differences between different groups, it's because people have different constraints
... that's what guides a certain way of answering a problem
... THese constraints are real
... That's why shims are useful

<ben_thatmustbeme> nothing stops a person from having multiple feeds, one for "boxing" that other boxing fans could follow

bblfish: Machine readable forms is interesting
... Translation between documents to machien readable data could work. Could be anote

<Loqi> Hhalpin made 1 edit to Socialwg

<sandro> ben_thatmustbeme, I don't think that's what that story is saying

<Loqi> Hhalpin made 1 edit to Socialwg/2015-03-17-minutes

AnnB: Much of this work is imporant for workflow management for us, expertise location, answering questions, sharing information, network and so on
... It has the attributes that are discussed here, but has a different quality to it than facebook and public tools
... One day I anticipate we will use these technologies with our supply chain (outside firewall)

<ben_thatmustbeme> sandro, i don't think so, but we haven't really tackled groups at all yet, other than tagging multiple people in a "group". Its difficult to get anywhere without some single owner to a group who is in charge of the feed

AnnB: Also because we're so big, being interoperable is mandatory
... Not good enough to have a shim

<ben_thatmustbeme> sandro, more of thinking out loud of possibilities

<aaronpk> ben_thatmustbeme: that's what I was getting at with aggregators

AnnB: Too big a domain

<aaronpk> like how a bunch of us are in the "indienews" group if we subscribe to

sandro: we should try to design this all without shims

AnnB: try to avoid, it's a patch

<ben_thatmustbeme> aaronpk, maybe just aggregator from set list of people "joined" to the group, looking for anything in a certain category

melvster: my understanding of json-ld is that it's design to be extensible
... where is it not seen to be extensible?

sandro: json-ld is, it's when it's just json

<ben_thatmustbeme> audio is in and out for me, so I have to go my chat logs to fill in the spaces

melvster: is that what the group is doing?

<aaronpk> ben_thatmustbeme, there are a number of ways to do it, feel like taking a stab at writing up a version of how to do it like I did for the other stories?

Arnaud: this is always a weak spot. This is a selling point for json-ld

sandro: we need some warning that says if you ignore the LD part this is the trouble you'll have

bblfish: peopel have been arguing about this for a while

harry: arguement is that most evidence says most json develoeprs will drop the context and don't know/care about json-ld

<cwebber2> if the context is implicit, and you have *enough* stuff defined, most users don't need to look at it, until they hit extensibility

<cwebber2> hm

sandro: if you have a different @context you might not even check if you're used to a standard

<cwebber2> sandro: that is a good point though

harry: how stable are microformats terms?

aaronpk: pretty stable like h-entry and h-card because based on previous work

<elf-pavlik> hhalpin

<cwebber2> though I think we talked about you not being override the stuff set up in the context though in new contexts though right?


aaronpk: we're adding new properties like u-like-of

harry: if we are really worried about context being lost it might be useful to walk through that
... where does activity vocabulary replicate existing mf work

<elf-pavlik> issue-15

<trackbot> issue-15 -- AS2.0 Vocabulary in many ways duplicates and efforts -- closed


aaronpk: do we already have this doc?


aaronpk: that needs some work

harry: microformats people should look at that
... see if we can make them converge

<eprodrom> I'm lost

harry: james just made vocabulary to map to usecases
... only microformats community can really help with going into this in detail

<cwebber2> is thiat about mappings again?

<eprodrom> What are we talking about?

harry: so then you move stable microformats into core

<cwebber2> or about having microformats having contexts

<cwebber2> or something

<cwebber2> or we can close on this :)

Arnaud: we can close, feel free to keep discussing

<tantek> hard to converge with something that (nearly) no one's publicly implementing in practice

<cwebber2> I'll be participating in dinner remotely!

<cwebber2> jk

<tantek> also - not necessarily good to converge something practical with something more theoretical

<eprodrom> rhiaro_++

<bblfish> thank you amy

Arnaud: Reusme tomorrow at 0900

<Loqi> rhiaro_ has 72 karma

<eprodrom> Thanks for scribing

<cwebber2> rhiaro++

<Loqi> rhiaro has 73 karma

<eprodrom> Thanks for everything everyone

<cwebber2> rhiaro: great job on scribing

<cwebber2> and yes good work everyone

<aaronpk> methinks those countdowns are going to be in PDT

<wseltzer> trackbot, end meeting