Web Annotation Working Group Teleconference

<TimCole> PROPOSED RESOLUTION: Minutes of the last WG call are approved: https://www.w3.org/2016/06/24-annotation-minutes.html

<azaroth> +1

<Jacob> +1

<ivan> +1

<TimCole> scribenick: azaroth

RESOLUTION: Minutes of the last WG call are approved: https://www.w3.org/2016/06/24-annotation-minutes.html

CR Status

Ivan: We have two published specs, model and vocab, on Tuesday. Went out with no problems
... Had a hiccough with the TAG issue that wasn't properly handled, but we got closure on Wednesday, and have everything to get it published
... Have a reply this morning that the document is ready to go
... So don't expect any problems and it will be published on Tuesday next week

<Loqi> I added a countdown for 7/12 12:00am (#5867)

Ivan: Editors and Chairs will be on the hook for any issues that might come up, as I'll be on vacation
... Have some editorial feedback (Rob and Benjamin) already, but will deal with it after tuesday

TimCole: Any further comments?

Rob: Hooray! :)

Tim, Ivan: A very important milestone :)

Testing

TimCole: There was some notes from Shane, that he has closed his pull request

Benjamin: He had been asked to squash the commit history into a single commit
... He was going to try and squash down to only the important commits, and rename to be more consistent
... The reason to do it is to not have the master list expanded by many new commit messages that overwhelm other messages in the history
... Shane now cleaning up his local history and will send a new PR
... Hopefully doesn't restart the process, but it is blocked on Shane's work

TimCole: Trying to get local version up, ran into some issues. Also confused about spec-ops and w3c test environments
... Are they the same thing?

bigbluehat: He has write access to spec-ops, w3c are the gatekeepers with the PR requirements
... If not added there, then not part of the W3C test suite

TimCole: Relationship to our own annotation-test repo?

bigbluehat: Canonical source of where we keep the schema and tests, regardless of the infrastructure
... decided to do that to avoid getting blocked on WPT if there's other tools
... WPT can pull from it periodically, but regardless of how those politics go, we still have our schema for testing

TimCole: Test scripts are tailored for the runner?

bigbluehat: Yes, but could write code to handle those too.

TimCole: Have some runner scripts, as you can't test without a runner

bigbluehat: Best to keep them in web anno tests. If WPT fails, which doesn't seem likely just a little slowed, having our tests is useful to have in our own repo, so more tools can benefit
... they're not dependent on any run time, just descriptive docs

TimCole: This has been about the data model

<TimCole> scribenick: TimCole

azaroth: when and how can we get the annotations that ??? produces to test?

bigbluehat: Are you wanting a place to past these examples?

azaroth: yes, so that we can demonstrate end-to-end, albeit incomplete.

<bigbluehat> http://shane.spec-ops.io:8000/

azaroth: the sooner we get that done the sooner we get more implementations

bigbluehat: we can use Shane's server in the short term

ivan: in many respects from the CR point of view what counts is to get back an implementation report from ???
... in a format that Greg K. can process

azaroth: given Shane is not on the call and has to worry about the WPT process, can someone help with this first end-to-end

<azaroth> scribenick: azaroth

TimCole: It looks simple, but there were some difficulties
... the runner doesn't work in my install
... it gets hung making the manifest
... What OS are you using?

bigbluehat: Windows, and ran into compile trouble

TimCole: I just cloned the repo and it started okay, but it got stuck. If someone was able to get it running, and we had the japanese examples, we could then generate a report to send to Gregg

+1

scribe: Wondering if we need a place to save the annotations?

Rob: With the intent to save them and rerun with new tests, or a different feature matrix

Ivan: No analogy to it in other groups. No harm to storing them, closer to the tests would be good. Nice to have, not a requirement

TimCole: You were thinking of starting with the japanese demo?

Rob: Yes, did an implementation of a client for IIIF, could do the round trip test

TimCole: Should identify a folder in the Web Anno Test repo for storing annotations to test

<tbdinesh> kamazaki annotator does not seen to work... for me.

TimCole: Seems simple enough.

Rob: Separate per implementation

Dinesh: That implementation doesn't work for me. It just says preparing, it doesn't allow annoting the image

Rob: I tried only with the example images they have

Dinesh: Will try more and write a report

TimCole: I know shane has other things going on, I wonder if we need an implementation of the platform we can use?

bigbluehat: Need to know what Gregg needs for the report. Then just running AJV over it several times

TimCole: Let's try and get that going. Need implementations to get started with the tests as soon as we can

Ivan: Would be worth reaching out to people we know are planning to implement to get them rolling

TimCole: Need to give them something to work with. Europeana is anxious to start
... Need somewhere to put the annotations and how to run the tests.

ivan: Might be worth talking to the Pund.it people, what they plan to do
... They said they want to transition to the model, but don't remember about timing

<tbdinesh> (FYI: it throws this error after a while. will write to him. image-annotator?u=http://static.zerochan.net/Uchiha.Sasuke.full.678195.jpg:31 Uncaught ReferenceError: tiledesc is not defined )

Rob: Don't recall if they were clean on timing

Ivan: Europeana, Pundit, Hypothes.is, IIIF all good targets

TimCole: That gives us some work to do while Ivan is on vacation :)
... Benjamin, protocol testing?

bigbluehat: I've begun reorganizing the tests I have, and the tests for the tester, based on the PR that Rob and I went through with the plan to put in after CR
... Easier to test.

TimCole: Just editorial?

bigbluehat: Yes, just clearer and better structured
... the code in WPT can be run separately from the wpt-serve system
... server could be made standalone and generate its own reports, if we need to
... protocol tester project has come a long way, so intend to finish that out and have people use to test protocol implementaitons
... then finish my protocol implementation to let clients test against that

TimCole: Is it in github yet?

bigbluehat: Yes, in a weird spot at the moment
... a work in progress

<bigbluehat> https://github.com/Spec-Ops/web-platform-tests/pull/3

bigbluehat: in need of rebasing
...: tophat:

TimCole: thoughts on the issues with WPT PRs

bigbluehat: When you work with someone else's code, they have opinions about how it should all be done
... not sure how broken it is. Sounds like he knows what's needed, just needs to have the time to do it
... some of his commits are in front of my PR, so you can see what he was trying to curate

TimCole: For what you're doing and for the model, when do we go forwards? Only have a small percentage of the schemas written

bigbluehat: So long as the test structures don't change wildly, no need to wait
... should ensure he can work on things when he has time, and not pulled off in other directions
... How are you testing the scripts now?

TimCole: ajv command line

bigbluehat: Could write those up, and in lieu of WPT, could point at the readme

<tbdinesh> +1 bigbluehat

bigbluehat: or wrap the commands in a script

TimCole: Other thing we've been doing is saving incorrect annotations to make sure the schema catches them
... so can say annotation doesn't correctly implement as it leaves out a required key

ivan: same for the protocol. THere's a number of MUSTs that need to be checked
... the changes, should wait until Tuesday before committing

TimCole: We won't talk about them till after Tuesday :)

ivan: I won't be here, so I created a Release in GH. If someone can create a protocol release that would be good.
... the changes on the protocol spec are good, fully behind it. I ran into the same problems understanding the paging.
... Made me realize my toy implementation is buggy :)
... That's the real value of CR

TimCole: Both model and protocol for the next couple of weeks, we'll have something to point implementers to
... that's what we're shooting for?
... combination of documentation and a little client/server tester

bigbluehat: Ship everything as soon as possible

TimCole: Don't want to miss the window of interest
... where are we on vocab testing? Primarily the context doc?

ivan: Already have a test ... forgot to update the context and the new one works for the japanese demo

TimCole: A good data point

ivan: Not all the implementations will use the context. Most won't be RDF based. I know the guy is semantic, so he did some testing by putting the data into an RDF store
... That can be documented
... same will hold for Europeana

Rob: At least transform through RDF on the way in and out

ivan: vocab testing is really about that -- the context reflects the vocabulary the way it should. A bit shallow but good data points
... Need to record them.

<TimCole> scribenick: TimCole

azaroth: similart vocab data point
... recent updates to my implementation of protocol transforms both ways json-ld to rdf to json-ld via context and frames
... turtle to json-ld and vice versa
... I could extract code and make stand-alone as a demo of ability to do this

ivan: yes, this is good,

<azaroth> scribenick: azaroth

TimCole: Would be good to document it

Ivan: Do you have a feeling of the RDF terms we define are in use?

<TimCole> scribenick: TimColle

<TimCole> azaroth: client implementation is just images

<TimCole> ... so svg selector, uses format, textual body, created, no creator yet

<TimCole> ... all the core stuff, but not the text stuff

<TimCole> ivan: adding a single table that these are the terms that have been used by x implementation(s)

<TimCole> ... we asked each of the implementations to keep track

<TimCole> ... to help us keep track, nothing else than an HTML table

<TimCole> ... so when skos was defined it was always a question about actual usage of the vocabulary

<TimCole> azaroth: yes, I can generate such a table.

<bigbluehat> +1

<azaroth> scribenick: azaroth

TimCole: Anything else about vocab testing?

Ivan: can document usage and context mapping. That's what we said we'd do on the transition call

TimCole: What else today?

<tbdinesh> we have tried a bit. will see if i can put together a "team" :)

TimCole: Nothing else?
... should we invite the Japanese and europeana folks to one of our calls to discuss testing with them?
... assuming we have something for them to use

<bigbluehat> +1 to hearing from implementers we know about

<bigbluehat> takeshi++ for being awesome

<Loqi> takeshi has 1 karma

Rob: Hard for the combination of Europe and Japan!

<bigbluehat> takeshi++ for being way more awesome than 1 karma point!

<Loqi> takeshi has 2 karma

TimCole: By email then when we have something to show
... Can adjourn unless there's other business
... will make progress on the schemas. If anyone wants to help, please jump in

ivan: By the time I come back from vacation... ?

TimCole: Should at least have some test results :)

<bigbluehat> ivan: you've earned this vacation to be sure!

TimCole: When do you come back?

+1

ivan: 1st of August

takeshi: Checking the json-ld, and looked at the RFC for the profile URI
... have developed a profile URI in the protocol, and should provide the information to the RFC

<takeshi> https://tools.ietf.org/rfc/rfc7284.txt

<TimCole> scribenick: TimCole

azaroth: at TPAC last year we discussed with Heather whether it would be possible to register
... a profile without registering a media type
... so we could register a profile

<bigbluehat> there's a separate profile registry iirc

azaroth: we should do that

<bigbluehat> oh...right RFC7284 >_>

ivan: do we know how to do that

azaroth: not clear since no one has registered a profile URI yet

<bigbluehat> start now, I'd say

ivan: this worries me, having been through some of this before
... there is no way to register a fragment id propertly

<bigbluehat> ...which means its not an official thing yet

ivan: not fully up to date with ietf, but this is just an informational document by Markus

<bigbluehat> https://datatracker.ietf.org/doc/rfc7284/

<bigbluehat> full status ^^

ivan: I presume Rob was in contact with Markus on some of this
... so perhaps Rob should contact Markus directly to ascertain status

<bigbluehat> send it to me, I'll +1 it and send it back. promise

azaroth: in Example 3 of the document there is a sample registration submittal

ivan: with regard to media types there is a W3C agreement with ietf
... such that media type defined in W3C document gets properly registered
... not sure if it works the same way for profile

azaroth: will make an issue in git hub (PR milestone) to get in touch with ietf

<ivan> trackbot, end telcon