Web Annotation Working Group Teleconference

TimCole: Agenda for today

... anyone wants to add something?

Announcements

<azaroth> bjdmeest: And continued thanks, as you're often the scribe!

TimCole: WD's are published
... reviews are requested

ivan: a reference from the security review people is received
... there is a questionnaire that helps identifying common pitfalls

<ivan> security questionnaire: https://w3ctag.github.io/security-questionnaire/

ivan: anyone willing to go through that offline?
... something that should be done between now and next version of the WD
... so if security people will check this, we need to check this asap

azaroth: I am happy to work with whoever would work on that

TimCole: I'll send a question around

<TimCole> http://w3c.github.io/web-annotation/selector-note/index-respec.html

TimCole: other thing to mention: there is a first draft of a possible note on selectors and states
... please take a look at this draft
... we could add github issues as we need to

ivan: two things:
... 90% of the text was cut'n paste
... one good check would be to check I didn't forget anything
... so to check whether the text remains consistent
... someone else should do that
... second: I came up with a simple way of putting things in a fragment, differently than the previous proposal
... the last few days, I made a small javascript lib to parse the fragment url into a JSON object (see http://w3c.github.io/web-annotation/selector-note/converter/)
... also, there are 2 new issues on the model

TimCole: because the tight coupling between model and this note, we need to make sure everything remains consistent

ivan: that's my job :)

Minutes

shepazu: The next 3 months I will focus on testing, together with ShaneM
... to get concrete work done on the testing

TimCole: other volunteers will be welcome
... occasional updates from you guys are welcome

shepazu: of course
... at this point, every week, we should have a testing update, during this telcon

<Tbdinesh> We will be pitching in too.

ShaneM: What's the target timeframe for CR?

ivan: we will have a F2F in Berlin in mid may
... after that meeting, we should go to CR
... issues will come up
... but we virtually published a last call
... that's the idea

<TimCole> https://www.w3.org/annotation/wiki/Meetings/F2F_Berlin_2016

ivan: the charter ends the end of September
... we will not make a recommendation by then, but we can report that we can to PR
... so that we can extend the charter to go through PR and rec

<ShaneM> remember that charter extensions require AC review now

ivan: so we would like to start CR beginning of June, and our of CR at the beginning of September
... W3M can - I think - extend the charter without changes to finnish the rec

TimCole: Hopefully we can get an extension

<TimCole> PROPOSED RESOULTION: Minutes of the previous call are approved: https://www.w3.org/2016/03/25-annotation-minutes.html

TimCole: any comments on the previous minutes?

<azaroth> +1

RESOLUTION: Minutes of the previous call are approved: https://www.w3.org/2016/03/25-annotation-minutes.html

Meeting Around the time of TPAC

TimCole: we had a F2F at TPAC, but for the next TPAC in September, it's not clear we need a full F2F
... as we have Berlin F2F, and would go to PR and CR by September
... proposal would be to have a full F2F, a virtual F2F of a couple of hours, or not a F2F at all

shepazu: if we don't have a F2F at TPAC, we shouldn't have a virtual F2F either

<fjh> +1 to doug re avoiding TPAC conflicts with virtual meeting

shepazu: it's not viable wrt people traveling to and from TPAC
... I think it's viable to have a F2F at TPAC, for communication with other WGs and CGs

TimCole: How quickly do we have to decide?

shepazu: It's better to say yes and possibly cancel instead of the other way around

ivan: making a final decision within max a month would be a good idea
... we have to say something by +- 15th of April
... registration will be closed by then

azaroth: 2 things to change our minds:
... one is the F2F in berlin in half of May
... is that too late?

shepazu: no, seems the natural tipping point

azaroth: given the time frame moving towards CR, I'm not convinced to meet at TPAC, if we do need to meet, something has gone wrong
... my preference is to keep it open until May, with the expectation that we won't meet, except if something comes up

TimCole: I see value in F2F, partly to talk with other WGs about implementations etc.
... so wait until May seems good

ivan: one of the chairs needs to register
... please try to register so that the meeting would be on Thursday and Friday
... let's do that now, and decide on the F2F in Berlin

TimCole: any comments?

<TimCole> Proposed RESOLUTION: We will submit request to meet at TPAC, and decide if we want to back out at F2F in Berlin

<azaroth> +1

azaroth: I'm happy to register and filling the forms

<TimCole> +1

+1

<ivan> 0

<ShaneM> +0\

<Kyrce> +1

<bigbluehat> +0

<shepazu> +1

<takeshi> +1

shepazu: there may be stuff in digital publishing that may be happening at that time
... so that might be useful to talk with them

RESOLUTION: We will submit request to meet at TPAC, and decide if we want to back out at F2F in Berlin

F2F topics

TimCole: we have 6 weeks until Berlin
... we probably need to start thinking about topics and agenda

ivan: if we keep to our current schema, then, it's difficult to plan in advance, because the goal is to close all technical issues
... difficult to plan now, if we get a load of issues from, e.g., i18n, we have to take that into account
... also, we need to have our CR criteria and testing etc. finalized, to have a credible plan to go into CR
... for me, it's difficult to plan now
... maybe we won't have any open issues, and spend 1,5 day on testing
... everything else has a lower priority

<ShaneM> If the agenda is testing focused, and assuming I get my board's approval, then I could probably attend the meeting in Berlin

TimCole: do we need to check our charter?

ivan: yes, when we go to CR, but that's the kind of thing we can do offline as well, we don't need the F2F for this
... getting the issues closed and CR verified are top priorities, everything else comes after that
... to come up with the report of what we accomplished etc., is chair work, and can be done offline

<TimCole> https://www.w3.org/annotation/wiki/Meetings/F2F_Berlin_2016

TimCole: let's not forget that, and plan some minutes for that
... also, please register anyone who will come
... The timeline for notes is not with a deadline, so we can postpone those (e.g., about HTML serialization)

Conformance

<TimCole> https://github.com/w3c/web-annotation/issues/165

TimCole: issue about conformance, there is a proposal
... this also was related to 'levels' of conformance, but we rejected that

ivan: there should be a way for an implementation to say that it does not deal with a specific media type
... so selection on those media types cannot be done by those implementations
... It can be as 'simple' as having a table of media types with their selectors
... and an extra table per implementation to show which media types it can be handle

<ShaneM> That makes sense to me

ivan: it's only the selector that created problems, not the rest of the model
... the model must me implemented fully by every implementation

azaroth: one thing I thought about is about the fragment selector
... the table of fragment selectors is not a normative requirement, bu maybe we should make it a bit more normative somehow
... so a client that can work with image/* Must implement the media fragment-syntax

shepazu: we could say that a Text class needs to do these selectors
... and these are not exclusive

<azaroth> +1 to shepazu

shepazu: it's just acknowledging the extension model

TimCole: first: do we have a template of an existing rec
... second: we talked about clients that create annotations, annotation repository, annotation consumers, annotation servers, agents that republish annotations
... it seems to me that will also be a factor in how conformance is described

shepazu: I agree, an ecosystem should consider all these things
... but I don't want us to get too ambitious with the testing
... I want to scope it to specifics that are called out in the existing specs
... e.g., we want to test the protocol (cfr. LDP)
... we could certain elements of the data model
... we could test the structure
... that the right objects have the properties of the right type, etc.
... we could also test to make sure selectors are interoperable
... I propose to leave that last one out of scope
... because we don't explicitly talk that any given UA will return exactly the same output for any selection

takeshi: about fragments selector conformance testing
... about plain text
... e.g., text would be encoded in an encoding system
... we haven't specified which encoding should be used
... I think it's important for conformance testing

<ShaneM> I dont think that encoding is a real problem. It is part of the headers (accept-encoding)

ivan: encodings is a very specific issue to be taken into account for text selector
... question is about specification about specific selector
... a specific issue is important

<azaroth> +1 to raising encoding issue, thanks Takeshi!

ivan: now, about the categories of the implementations:
... let's not over-formalize ourselves
... we had in the past implementations that implemented certain parts, but not the other parts
... and that's fine
... it's not up to us to decide on that
... what we really have to be careful about, is that each feature needs to be implemented by 2 different implementations
... having all these very formal categories, would drive us to a very formal way of conformance testing
... that may become very complicated

TimCole: there has to be some recognition

ivan: the issue about media types may be - where we have to stop - is to have a very clear description in the text

<shepazu> +1

ivan: These are the selectors that are relevant to these media types
... to make clear that text/* media types are handled by selector A and B, and are thus relevant to me

ivan: that as and addendum will guide us to have test suites on all those categories

azaroth: so, we start with syntactic testing, conformance with JSON, correct properties, etc. (cfr JSON schema)
... seems low-hanging fruit
... less than a day of work
... then, we can start with protocol
... LDP had some good patterns
... also, sending all possible annotation forms to a server, and a server that sends all possible annotations to a client
... at that point, we can see what else we can do

shepazu: we agree about semantic vs syntactic testing
... I like the framing
... what about the OA CG testing?

azaroth: in the IDPF work, we created a JSON schema for the CG open annotation model

<azaroth> http://www.idpf.org/epub/oa/#h.b2nk2onxjepf

azaroth: at the bottom of the spec, there is the JSON schema

ivan: IDPF people will want to take the new model, so that's important

shepazu: good that we can start from the schema, thanks rob!

<ivan> http://www.idpf.org/epub/oa/schema/oa-epub-schema.json : the real schema

azaroth: It's on my list
... by the end of next week, I hope to have done that

<ShaneM> We have some JSON-LD testing code sitting around that needs polishing, but hoping it will work well for syntactic testing in the web platform test framework.

TimCole: that's a good consensus about an approach for the conformance
... it would be nice to have some more examples

<azaroth> Regrets for next week, yes

TimCole: for next Friday, both Rob and Ivan aren't here
... I have on the list: HTML serialization (which would be a note)
... without some guidance, there won't be much uptake

shepazu: I'm interested, I'll be on the call next week
... I think we can also talk about testing

<azaroth> Sorry, need to disappear. Bye all!

TimCole: ok, so the call of next week will be about testing and HTML
... other ideas or examples are welcome
... [adjourn]

<bigbluehat> +1

<bigbluehat> bye all!

<ivan> trackbot, end telcon