Web Annotation Working Group Teleconference

<ivan> Agenda: https://lists.w3.org/Archives/Public/public-annotation/2016Apr/0104.html

<azaroth> Chair: Tim_Cole, Rob_Sanderson

<scribe> ScribeNick: fjh

Scribe Selection, Agenda Review, Announcements

azaroth: first 30 min for agenda followed by f2f and issue for 2nd half
... had call with PING yesterday, will recap
... no changes to agenda noted
... no announcements, lets discuss iAnnotate under F2F topic

Minutes approval

<azaroth> PROPOSED RESOLUTION: Minutes of the previous call are approved: https://www.w3.org/2016/04/22-annotation-minutes.html

<ivan> +1

RESOLUTION: Minutes of the previous call are approved: https://www.w3.org/2016/04/22-annotation-minutes.html

Testing

azaroth: need to clarify time frames and tasks to be done
... and who can help
... how to break vocabulary into smaller tests, need proposal and recipe for implementers to participate and contribute new tests
... also who will test implementations when working with other groups

ShaneM: purpose of w3c testing is to verify for each feature there are at least 2 implementations that support it

actually it is up to group to set the bar

ShaneM: not to certify implementations

<ivan> +1 to ShaneM

fjh: +1 to ShaneM re not certifying

shepazu: need to run tests on real implementations

<bigbluehat> in this case don't we test the output of the implementations? their JSON(-LD) serializations of whatever they're storing?

ShaneM: would like to test output of implementations but also need to test JSON schema, data model representation

<azaroth> +1 to "representation of the data model" :) that's a nice way to frame it (IMO)

ShaneM: not to make sure correct
... test JSON schema against canned input
... 2 kinds of testing
... testing against implementations - need to find them, need to get JSON schema completed

<Zakim> azaroth, you wanted to ask about /other/ means?

azaroth: what if we create tool for testing but wouldn’t suffice to meet w3c requirements

shepazu: could create annotation bot, e.g. one that finds typos and annotates them

<bigbluehat> PaoloCiccarese: yes. you can write it to allow extra properties

PaoloCiccarese: how to we test JSON schema if we have additional properties

<bigbluehat> it does that by default, in fact

shepazu: can we test the JSON schema and ignore additional material

<TimCole> So, do we need samples of resources that are meant to be annotated a particular way, and then see if implementations create json-ld that describes the annotation?

azaroth: answer is yes

ShaneM: JSON does not know about prefixes, so need to normalize input

bigbluehat: using keynames in anything in profile, more constrained than JSON-LD, require context and ranges for values
... so JSON schema validation should be ok

azaroth: agree

<bigbluehat> https://www.w3.org/TR/annotation-model/#index-of-json-keys

azaroth: if property constraints to false then anything not specified is ok

<bigbluehat> also: https://www.w3.org/TR/annotation-model/#model "MUST have 1 or more @context and http://www.w3.org/ns/anno.jsonld MUST be one of them. If there is only one value, then it MUST be provided as a string."

PaoloCiccarese: in general dealing with JSON-LD in JSON environments prefix is handled inconsistently , expanded or not, so what are we going to do

bigbluehat: can use vocabulary how you see fit, for annotation model then have to use key names

shepazu: gkellog mentioned framing for normalization

<ShaneM> unfortunately framing is not mature enough for us to reply upon at this time

<azaroth> json schema validation with additionalProperties of false: https://tools.ietf.org/html/draft-fge-json-schema-validation-00#page-13

<ShaneM> at least that is my position

shepazu: we might want to get him on a call

<azaroth> And the OA json-schema for IDPF: http://www.idpf.org/epub/oa/#h.b2nk2onxjepf

bigbluehat: ... validate against vocabulary, if web annotation also validate against JSON schema

ShaneM: let’s get something working first, if too constraining reduce the constraints

shepazu: need to make sure this works within W3C testing framework, however many if not most implementations cannot use web framework
... so will need manual validation
... should focus on that first

<PaoloCiccarese> FYI: In Annotopia I use Framing and then I do custom validation, which I assume can be swapped with JSON Schema pretty easily as long as the framing produces an agreed upon output

shepazu: start with a validator

ShaneM: disagree, doing assertion based tests

shepazu: mean doing tests manually, possibly using a web form to enter inputs and get results

ShaneM: web test environment supports that directly, so don’t need to reimplement

shepazu: didn’t realize that, ok
... lets talk offline to be clear

ivan: JSON framing not implemented widely so not sure implementions can use, algorithm defined, but not a REC
... JSON schema should not be normative

<ShaneM> +1 to not making it normative.

ivan: if normative, then need to make sure absolutely consistent with RDF vocabulary

<azaroth> +1 to Ivan

ivan: if inconsistency between schema and vocabulary, vocabulary wins

ivan: need someone to work with on schema

PaoloCiccarese: framing is essential if you do RDF, no way around it
... but starting point is schema validation, can work on first step of pipeline, can create some RDF test case
... happy to be involved

<bigbluehat> 2 things you should be able to check: "my RDF uses the Annotation Vocabulary correctly" and "my output JSON format is valid Web Annotation Data Model JSON Serialization"

<bigbluehat> PaoloCiccarese: you may not need to do both (afaik) ^^

azaroth: framing proposal is not that every implementation has to do it, but that test generator has option to use it

ivan: ok with that, Gregg can do it

azaroth: i can work on JSON schema side

<Zakim> tbdinesh, you wanted to comment on process for suggesting test cases and validating tests do what they say

<bigbluehat> https://github.com/bigbluehat/testing-json-ld <-- this thing

tbdinesh: can make more test cases, what is process for doing this
... what if I want to test multiple targets

<bigbluehat> these https://github.com/BigBlueHat/testing-json-ld/tree/master/web-annotation/tests

<bigbluehat> happy to move these into the w3c space on GH

tbdinesh: need wiki for tests, so we can look at them, know what they are doing

ivan: agree

<bigbluehat> PRs welcome ^_^

tbdinesh: need list of needed tests

bigbluehat: instead of one big JSON schema, use one per MUST, using defaults in JSON schema, see link above

<bigbluehat> https://github.com/BigBlueHat/testing-json-ld/blob/master/web-annotation/tests/verify-target-present.json

bigbluehat: need to decide some types, string etc
... deal with arrays and streams appropriately
... different approach than giant schema

<bigbluehat> https://github.com/BigBlueHat/testing-json-ld#screenshot

<azaroth> In my experience, many small tests is better

<azaroth> as the mega schema will stop as soon as it hits the first error

<azaroth> Also there's no distinction between error and warning

<azaroth> (MUST vs SHOULD)

<azaroth> so you stop after the first warning even

<tbdinesh> for example, for motivation renarration i need to define new motivation and then its input validation for those uses

bigbluehat: uses quads, relies on human involvement, but a starting point

<ShaneM> azaroth: yes. that's how the WPT works

<ShaneM> atomic tests are key

Agenda for F2F

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

azaroth: continue this on next call
... comments on agenda, proposals?
... how many observers do we have?

ivan: suggest we change agenda, focus on topics related to going to CR
... if time, include others
... remove client-side APIs, search, robust anchoring etc
... need to close all issues to go to CR, be clear on testing strategy, before going to CR

<azaroth> +1

ivan: this is first priority

TimCole: agree with Ivan, however suggest meeting in 3rds, 1st afternoon on testing, morning on issues, put otther items we want to do before charter expires as last third
... if we don’t have time then we can slip them to later calls

+1 to TimCole

TimCole: we should break down testing to sub-topics, schema, framing, implementations etc

<azaroth> and +1 from me too

+1 to ivan’s suggestion to prioritize

azaroth: will revise agenda, we can discuss next week

dwhly: not much new to report on iAnnotate, planning continues, please attend and remember to register

<azaroth> Can you drop a link to the registration page?

dwhly: lots of participants, about 120, increasing daily, max will be 150

<tbdinesh> iannotate.org

dwhly: remember Sat 1 day developers meeting, sign up for that separately
... working on panel on harrassment and page owner consent over annotation, should be interesting
... Genius will be there

<shepazu> +1

dwhly: +1 to Ivan, however perhaps have some time in F2F to talk about consent, or on a call before, then input into panel discussion
... at iAnnotate, not definitive statement, but suggestions or additional information

Issues

issue-195

<azaroth> proposal is: https://github.com/w3c/web-annotation/issues/195#issuecomment-213490285

dwhly: in progress

TimCole: Privacy in CR documents takes precedence over future privacy work
... lets look at issue 195

azaroth: selectors and sub-selectors, merged using refined by to allow state or selector, so now question can have both, proposal is yes

<TimCole> Proposed Recommendation: Accept proposal and close issue #195

<azaroth> PROPOSED RESOLUTION: Allow a State to be refined by a Selector.

azaroth: had support for proposal from Ivan and Jacob, no concerns from anyone

<TimCole> +1

<ivan> +1

<azaroth> +1

+1

<PaoloCiccarese> +1

<ShaneM> +0

<bigbluehat> +1

RESOLUTION: Allow a State to be refined by a Selector.

<tbdinesh> +1

<TimCole> https://github.com/w3c/web-annotation/issues/205

TimCole: should we close

<azaroth> Agree it's incomplete

ivan: document incomplete now, allow two selectors or states on top level, spec silent on meaning
... refinement covers various use cases
... two means conjunction
... don’t really like this, could disallow
... my preference

TimCole: take to github

ivan: we had disagreement so we need to decide

<TimCole> https://github.com/w3c/web-annotation/issues/206

TimCole: let’s take it next week

ivan: text position selector is under specified as noted in the issue
... should we say anything in model about encoding

azaroth: we don’t, agree spec is incomplete, as Takeshi noted as well

ivan: if we use HTML5 then encoding is defined

<ShaneM> technically it is part of the wrapper

ivan: cannot have our own definition that conflicts with HTML5
... wrapper

azaroth: Takeshi made web page of various languages and frameworks
... on how dealing with characters

TimCole: Rob, Benjamin, Paolo willing to help with schema issue
... anybody else?

<ShaneM> me me me

<shepazu> ShaneM

<shepazu> shepazu

TimCole: ask that group to get something started

ivan: to speed up if others could look at 205, 206, 191 and give opinion
... please work on the list, before call, so we can resolve them. its been 3 weeks
... these are technical?

<ivan> trackbot, end telcon