From W3C Wiki

05 Jan 2016


See also: IRC log


Arnaud, aaronpk, rhiaro, sandro, wilkie, jasnell, ben_thatmustbeme, tantek, tsyesika, wseltzer, akuckartz, cwebber


Summary of Resolutions

  1. Approve minutes from 2015-12-15
  2. Publish Post Type Discovery as a FPWD
  3. Close issue 261 without further action
  4. Close 276, changing to a MUST

<scribe> scribenick: rhiaro

<akuckartz> No audio yet

<jasnell_> running late, be right there

Arnaud: welcome back, happy new year!

approval of minutes from 2015-12-15

PROPOSAL: Approve minutes from 2015-12-15

Arnaud: any objections? anyone looked at them?


<akuckartz> agenda is empty?!

sandro: They contain approval to go to fpwd for two drafts

<tantek> +1

Arnaud: hearing no objections, approved

RESOLUTION: Approve minutes from 2015-12-15

scribe: Reminder, face to face proposed march 16th
... Page set up, please specify your expected attendance
... As usual, if you're not sure please say so


scribe: Better than saying nothing
... As you remember I sent regrets for the last few meetings, but noticed that agenda ended at the last call with the post-type discovery draft going to fpwd, and that was postponed to this week

Post type discover draft

tantek: We've had this up for a while, a few issues and some resolved
... a few implementaitons of parts of it
... I think it's in public shape for fpwd
... So I request the group consider that


tantek: If there are specific issues that must be resolved before we take it to fpwd pleases say what the issue number is
... Other than that, general feedback, is it ready or not. I think it's ready.

Arnaud: Comments?
... Need more time?

<wilkie> I think it is fine

Arnaud: If you need more time you could object and ask for time
... Been on the agenda for a while

<tantek> PROPOSED: Publish Post Type Discovery as a FPWD

Arnaud: Don't see any reason to delay further

<azaroth> I haven't implemented it, but have read it and see no reason to delay FPWD

<ben_thatmustbeme> +1

<azaroth> +1

Arnaud: Please vote

<wilkie> +1

<Loqi> Tantekelik made 1 edit to Socialwg/2016-01-05

<aaronpk> +1

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

Arnaud: For FPWD we don't need implementation, to clarify

<ben_thatmustbeme> there are 10 ppl on the call

<jasnell_> +0


<sandro> +0

scribe: I think we can call it resolved. Last call?

RESOLUTION: Publish Post Type Discovery as a FPWD

scribe: Sandro, you'll have to send the transition request

sandro: is it all under respec? or need to be reformatted?

tantek: I probably have some respec reformatting and all the pubrules checking to do

sandro: could you do that in the next day or so? then we could do all three documents on thursday

<Loqi> I added a countdown for 1/7 12:00am (#5785)

sandro: Just less work to do three at once

tantek: I'll see what I can do, but don't hold up the others

Arnaud: key part is, tantek you have the ball, once you're done let sandro know
... Sandro you can still request the transition in th emeantime


UNKNOWN_SPEAKER: Ready to go to CR? Still a few issues left, some editorial
... 276 in particular postponed because someone was missing
... James?


jasnell: want to ensure that vocabulary document has examples of all features
... Examples there don't illustrate everything
... I can add those, but PRs would be helpful
... tantek opened one about it needing a section on privacy implications, current working draft should have that
... would like everyoen to take a look and see ifthat's adequate
... Other editorial is 278, updating acknowledgements section, which I will do


jasnell: Two substantive issues, 261 and 276


<cwebber2> hi

jasnell: 261 talks about lifecycle and naming conventions for extensions, came up at last f2f
... So far no proposal put forward
... 276 might be easier after 261

Arnaud: cwebber wants to talk about 261?

jasnell: 261 was a conversation between sandro and tantek at the f2f
... do you want to discuss?

<cwebber2> huh, do I?

tantek: I captured because sandro felt strongly about it
... I'll leave to sandro

<Arnaud> I may be mixing issues and people :)


<cwebber2> oh... Arnaud I think it's probably important but I don't have much to say at the moment

<tantek> "Need lifecycle and naming convention guidance for extensions to help possible future convergence and standardization "

jasnell: goes back to whether we need to capture naming convention or registry for extensions
... for future standardisation
... and whether that needs to be documented in the core draft

sandro: I don't think so. I mean.. I have been thinking about this a lot and I don't think there's anything we can do right now that will really help

jasnell: so close without action?

sandro: yep

<tantek> I'm ok with that

PROPOSED: Close issue 261 without further action

<tantek> +1

<jasnell_> +1

<sandro> +1

Arnaud: please vote

<azaroth> +1

<akuckartz> +0

<wilkie> +1


<ben_thatmustbeme> +1

<dromasca> +1

<cwebber2> +0

RESOLUTION: Close issue 261 without further action

jasnell: next is 276, require valid as2 implementations to use vocabulary where applicable
... So when we're using an extension, say as a type, eg. Person
... THe implementor should also use the corresponding core vocabulary type
... So the type field would list schema:Person and as:Person

<jasnell_> "type": ["schema:Person", "Person"]

<tantek> that must seems reasonable

jasnell: That is so that anyone who doesn't understand the extesion can still make use of that by falling back to core vocab term
... Recommendation is for interop
... Spec says should, issue says make it a MUST

<tantek> as in, prevent extension makers/users from hijacking core spec semantics

<tantek> and claim compliance

jasnell: Personally I think should is fine
... I know in the comment thread others said SHOULD is okay
... but we need to get clarification
... I'd like to propose keeping should
... ie close without further action

<tantek> I'm leaning more towards MUST

<tantek> What are the objections to MUST?

<tantek> I think it makes the spec stronger FWIW

if you have a really good reason not to use AS2, why are you using AS2 at all?

Arnaud: use cases to argue one or the other?

<tantek> MUST tends to be better for interop

jasnell: use case is interop

<sandro> tantek, can you update to explain which of the two drafts it links to is the real one?

<ben_thatmustbeme> must means extensions have to have two types, which i find annoying personally

<tantek> I can't think of any good reason to allow foo:person and allow them to NOT say 'Person"

jasnell: if someone lists only, someone who doesn't understand might not know there is a core type to correlate
... tantek, want to speak?

tantek: I think that when in doubt standards tend to work better when conservative, MUST is more consvertaive. Tightens up interop.
... If people feel ambivalent about it, or no strong opinion, we should go with MUST

<cwebber2> I'm -0 on MUST

tantek: Unless there's an objection to must
... More MUST = better interop

<cwebber2> +1 on SHOULD

tantek: Curious about use case for SHOULD

<jasnell_> I have no objection to must but think the should is ok

tantek: I think that's where burden of proof is

<Zakim> azaroth, you wanted to be in favor of MUST

tantek: shouldb e MUST by default

azaroth: also in favour of MUST for similar reasons
... Should be explicit about what is required, and promoting classes and properties that the standard requires is just part of every day life. To say that you not-must use the features when you don't want to just makes it more difficult for everyone to implement

ben_thatmustbeme: My only concern is that extensions are forced to have multiple types
... if they want their own types. Annoying as a developer. Can live with, but is a nuisance

<jasnell_> lets perhaps split the question: show of hands in favor of must vs show of hands in favor of keeping it as should

<tantek> it's not for all extensions

<tantek> it's only "where applicable "

<cwebber2> o/ should

<wilkie> "MUST" to promote interop seems strange. do you add MUST for every new thing? hmm.

<tantek> ben_thatmustbeme: note: "where applicable "

<tantek> poll should vs must?

Arnaud: straw poll

<tantek> wilkie: yes, MUST is how standards indicate their core requirements

<tantek> +MUST

<wilkie> I know what MUST means

<Arnaud> STRAWPOLL: a) keep it a SHOULD, b) make it a MUST

<azaroth> Right. Extensions that don't conflict with the core wouldn't have multiple types

<tantek> b

<Arnaud> a: x, b: y

<jasnell_> a) +1 b) +0

Arnaud: +/-/0 responses to both in one line

<azaroth> a: -0, b: +1

a) +0 b) +1

<tantek> a: -0, b: +1

<sandro> a) -0 b) +0

<akuckartz> a: +0.5 b: +0.5 :-)

<ben_thatmustbeme> a: +1 b: 0

<wilkie> a) +1 b) +0

<ben_thatmustbeme> no real objections either way

Arnaud: argh...

<cwebber2> a)+1 b) -0

jasnell: I would wager to say that everyone could live with MUST

<ben_thatmustbeme> I can live with MUST

jasnell: even if folks are okay with should

<ben_thatmustbeme> it certainly does improve interop

sandro: only negative on must is chris with -0

<cwebber2> I can live with MUST, though I think it's strange to have it that strong

sandro: chris wanst to voice?

<cwebber2> but I can live with it

Arnaud: I'll cast a vote to try to make thins different

<cwebber2> I think that it would be strange to point to an AS2 implementation and point at a term and say "you didn't include our term, you aren't compliant"

<cwebber2> thus I think SHOULD makes more sense

Arnaud: I'm sensitive to tantek's arguement that the more MUST we have the more interop we have

<cwebber2> that's my voice

<cwebber2> but, I think it doesn't make or break things

<Arnaud> PROPOSED: Close 276, changing to a MUST

<jasnell_> +1


<tantek> +1

<azaroth> +1

<wilkie> cwebber2: that's essentially exactly how I feel

<wilkie> +0

<rhiaro> cwebber2 only if there's an applicable type, right?

<akuckartz> +0.5

<cwebber2> however I'm also happy to have a resolution

<rhiaro> so if there *is* an applicable type, why wouldn't you use it?

<sandro> +0

<ben_thatmustbeme> 0

<tantek> cwebber2: that's exactly what the MUST is for, so an implementer can't just make up their own duplicate versions of the terms in AS2 Core and claim compliance without using AS2 terms in actuality.

<rhiaro> if there isn't an applicable as2 type, no big deal

<jasnell_> thank you cwebber2 .. yeah, it's only applicable if there's an overlapping type in the core

<cwebber2> 0

<cwebber2> note that going through trains makes it hard to vote

Arnaud: we don't have to rush if you want to discuss further

<jasnell_> understood cwebber2

<cwebber2> I'm 0 becaus I am fine with closing with MUST

<cwebber2> if it means closing it

<cwebber2> I don't think it's worth keeping open

jasnell: I can add language to clarify that this is really when there are overlapping terms
... It's not saying you MUST use one every time, only when there's overlap

<cwebber2> jasnell_, that is fine with me

<tantek> Also, making it a MUST will help force implementers to reveal any specific use-cases to the contrary (i.e. for a SHOULD)

<cwebber2> and helps

Arnaud: I think we can close it

<cwebber2> +0 with that clarification

RESOLUTION: Close 276, changing to a MUST

Arnaud: ready for CR then?

<cwebber2> er

<cwebber2> I should say

jasnell: Will get a new wd out this week, and go from there

<cwebber2> going through tunnels makes it hard to vote :)

<cwebber2> mountain tunnels aren't exactly great for tethering :)

Arnaud: can we agree to go to CR pendng james's edits?

<tantek> ok will add an issue

<cwebber2> +1 on CR pending edits

sandro: we can't go to CR without figuring out exit criteria, and we should probably ?? tests

Arnaud: for now we'll let james edit, get it in shape, then tackle those issues next week

<tantek> exit criteria:

Arnaud: Maybe now?

<tantek> just created

<tantek> for tracking purposes

Arnaud: tantek, you're pointing to an issue..
... proposal?

tantek: just wanted to capture what sandro said

Arnaud: usually exit criteria means there are two implementations of every feature
... doesn't mean they have to be in the same implementation

sandro: need to clarify what it means to be an implementation
... producer, consumer..
... Vague sense we had consensus at face to face, but not clear

Arnaud: we could talk more now or put at the end of agenda for today
... Mostly what's left today is status updates

<Zakim> jasnell_, you wanted to talk about test suite


jasnell: as far as test suite is concerned, I did create a new github repo to start collecting sample activitystreams documents
... PRs to add samples to these would be helpful, I'll add as I have time (which I haven't)
... will continue to add myself too
... intent of this is to give us a corpus of test documents that illustrate every feature of the vocabulary
... so that implementations can go test to make sure that they are compliant

Arnaud: so what's important is that when we publish the cr draft is we have a pointer that leads people to a page where they can understand where to start and how to test their implementation

sandro: and where to post test results

Arnaud: okay if it's not complete, but we at least need a pointer in the document

sandro: gonna be some press around it saying this is our call for implementations
... so good to have instructions (complete) on day one
... strongly suggest we have test suite done

<akuckartz> Samples of non-AS2 documents will also be helpful

sandro: we can fix it later
... but should have it ready to go and copmlete before CR
... Depending on whether we're trying to sneak through or get something good here

tantek: from my understanding we don't actually need to have a complete test suite to enter CR, even to have a test suite at all, but the more we do have the more real it looks and the more implementable it looks
... so this is going to have to be a judgement call
... strictly from a process perspective we can enter CR as long as we have agreed on exit criteria
... that meet w3c exit criteria expectations
... The biggest challenge with the test suite is not to have one that's complete but to have a test suite that lets us have some common understanidng
... last time we got into a heavy discussion about the test suite there were different opinions about what that meant
... if we have to figure this out, the biggest challenge is how to say this implementation does something with the vocabulary that makes it necessary to be in the spec
... If all we have is tests that converts a stream between syntaxes, all it's testing is syntax parsing, not justifying anything in the vocabulary
... That's the biggest problem of considering whether a feature is 'implemented\
... That's the biggest challeng

<sandro> +1 tantek -- the tests should really show the vocabulary being used

tantek: Don't have a proposal, more just want to raise this issue as an area of past disagreemtn that we need to resolve on
... Also an area that evan and myself have disagreed on in the past
... other opinions on this, people should speak up

<akuckartz> tantek++

<Loqi> tantek has 267 karma

tantek: say what they think a featur emeans, and what testing that feature means

<azaroth> +1 to tantek, and +1 to not delaying -entering- CR before having the test suite

Arnaud: i recall, two aspects
... testing producers, can do with a validator
... testing consumers, have a suite of documents that james was talking about

<ben_thatmustbeme> a validator should certainly be a very minimum for our "test suite"

Arnaud: How you test wehther the application really understands it is still open ended I believe

sandro: I agree with what tantek just said
... i think cwebber2 and I were talking after the f2f... understanding I came to is that we should have a validator that at least gives a simple human readable description of what the machine readable syntax says
... Producers check their output against this validator
... And they send us an email saying all the terms they used and checked against the validator
... and we keep a table of all the terms they said that about
... and Consumers, need to use our suite of documents to tell us which terms their consumer can consume
... and their good faith assertion that it consumes them properly and understands what they mean
... and they might want to test that side by side with a validator. Look at a document in our validator and their thing, and say if they agree they make sense in the same way
... So we have a list of terms people produce and consume and believ ethey do so correctly because it aligns with our validator

Arnaud: comment?

<ben_thatmustbeme> sandro++

<Loqi> sandro has 26 karma

tantek: good start

<akuckartz> sandro++

<Loqi> sandro has 27 karma

tantek: biggest challenge is validating that a vocabulary is being used correctly
... is one thing, but part of the purpose of exiting CR is saying
... these feature are *doing something*
... if that something is abstrac tor concrete... by concrete I mean in a UI

<tantek> the challenge is whether the abstract vocabulary does something specifically in the UX

sandro: I'd like the reports from people not just to be a checklist of terms, but a paragraph saying what they're doin gwith it
... Hope that would help get the ball rolling if people see what others ar edoing

Arnaud: sandro since you seem opinionated about what we are to do, can you make a proposal?
... By next week

sandro: Sure.. the hard work is going to be having the test suite and validator and collecting results
... I think we need volunteers to do all those things
... But I'm happy to write that proposal

Arnaud: Quite a bit of work, not sure if we can afford to wait for it all to be done for CR
... Would be in favour of adding framework set up, publish CR with pointer, and keep on working on it

sandro: be nice to have the beginnings of a test suite and hopefully a partial validator
... but could say coming soon

tantek: this is where if we are specific in the exit criteria and we say that we say there are two or more implemenations of every feature where an implemenation means they are producing/consuming syntax correctly, and also producing some result that is unique to that vocabulary term as compared to other terms
... some language like that to put strenght into CR exit critiera
... that buys us time to create broader test framework
... we don't even need to have it partially done before CR
... I don't want to delay CR based on test suite

sandro: I'm not going to stand on the road over this, but If eel like in terms of developer engagement, but I'd rather hit them with one.. here's a time to look at AS2, and have the ducks in a row
... meaning there's a validator, examples
... and here's one or two implementations from inside WG
... as a developer I'd have a much more positive reaction
... if it all says coming soon, I could ignore

Arnaud: I agree we don't want to have nothing, but there's a middle ground
... We have a basic framework set up, and say as we move forward we'll add

sandro: has to be enough that i can start to play right now
... has to be a working validator for at least some stuff
... and a working set of info documents
... if I can't write code today and get in a feedback loop with the WG

tantek: I sympathise with what sandro is saying from a developer perspective. In the braoder view, AS2 has been in the discussion for years before w3c etc

<ben_thatmustbeme> i'd agree with sandro, as a developer i would want a validator and set of documents

tantek: Some level of built up expectation that we have inherited
... This is not brand new, it's been maturing for years
... Expectation that more of these tools and tests exist
... Absence may give people the wrong impression
... The fact that as2 has been going on for years before w3c has touched it. People have heard of it, already have an idea of what it means
... There's been a lot of work, we're not just rubber stamping, we need to show developers where the work is
... Devleopers can say 'okay now I get why this is real now'
... We have to figure something out, I don't know how

Arnaud: i don't think we're far away, just a matter of finding the sweet spot so we can got to CR without waiting for too long
... Suggest we leave it at this for today, sandro will put proposal together
... Feel free to discuss between now and next week
... Next week we can see what it will take
... See who volunteers to help out etc

Status updates

UNKNOWN_SPEAKER: Docs approved for publication last week. Sandro?

sandro: I think it's in my lap at this point. We can't use echidna for fpwd, so a bunch of manual steps, which can only be done on thursday

<Loqi> I added a countdown for 1/7 12:00am (#5786)

Arnaud: we got approval from wendy

sandro: I have to convert to static html and run a bunch of tests

Arnaud: maybe thursday, if not tuesay next week

<Loqi> Ok, I cancelled it

Arnaud: Reminder on giving everybody until 12th Jan to raise issues in ActivityPump and micropub
... Don't miss out
... Expectation to go to fpwd
... Finally, admin.. looking at tracker, a couple of issues
... when I looked at actions, baffled by list of open actions


Arnaud: Don't care to go through this now, but highly recommend that people look, several of you have open actions
... We should be able to close some of these
... Either they have become irrelevant or you can tell us what's stopping you
... But looks silly to have actions due over a year ago still open
... Please ahve a closer look and see where cleanup can be done?
... That leads us to the end of the call

<ben_thatmustbeme> we have actions for people no longer in the WG

Arnaud: Next call next week, resuming normal weekly callls
... Any other items?

sandro: on the open actions, some of mine are obsolete and we basically just stopped using the tracker
... should just mark as done? Others are more debateable?
... Shouldl people just mark them done or invovle the rest of the group?

Arnaud: personally I'm fine with people marking them as done
... There is a history of changes
... So if there's abuse, it can be dug up
... Meant to be a help to the group

sandro: mark as closed if it's not controversial, pending review if you want to talk about it

Arnaud: yes, then we can discuss on calls
... If there's something that really ought to be done, please do it!
... Any other business?
... Let's close on this

<akuckartz> bye!

Arnaud: Thanks!

<wilkie> rhiaro++

<azaroth> Bye all! :)

<wilkie> thanks all

<Loqi> rhiaro has 194 karma

<Arnaud> trackbot, end meeting