Web Annotation Working Group Teleconference

<fjh> trackbot, start telecon

<trackbot> Date: 24 June 2015

<scribe> scribenick: bjdmeest

Agenda Review, Scribe Selection, Announcements

<fjh> no announcements

Minutes Approval

<fjh> proposed RESOLUTION: Minutes from 17 June approved, http://www.w3.org/2015/06/17-annotation-minutes.html

RESOLUTION: Minutes from 17 June approved, http://www.w3.org/2015/06/17-annotation-minutes.html

<plh> http://testthewebforward.org/docs/

Testing

plh: we have done testing for years
... 3 years ago, we started to harmonize
... there is an open source project now
... where the WG can foster their tests
... primary objective: testing web browsers
... docs are available (writing, running, reviewing tests)

<plh> https://github.com/w3c/web-platform-tests/

plh: it is organized around git repos
... from HTML to Web Driver etc. huge lib available
... we have a dedicated system to review tests
... tests are run as much as possible automatically
... manual tests exist as well (e.g. pointer events), but we prioritize automatic testing
... goal is to test the web
... e.g, mozilla executes some of the provided tests automatically on every commit
... explorer as well, chrome is following, apple is next
... Tooling is all written in Python
... an auto pilot (in python) executes all tests sequentially

azaroth: [about ontologies and data models] is there any past experience about testing those?

plh: yes, but not in a cohesive way
... there has been some work about metadata, but there is a lot more variety of tools to use for testing
... e.g., LDP group used JAVA to write their own testing tools

azaroth: other relevant WG with experience?
... in the world of datamodel

ivan: we should talk to Gregg Kellogg
... he did stuff for RDFa, RDF, JSON-LD, CVS on the Web
... Gregg has some great tools, i.e., to generate reports

shepazu: we should try to integrate with the currently available testing framework
... that is beneficial for us and other WG
... tighter cohesion between specs
... this WG has tests for an Ontology and protocol
... that has precedence (XHR testing)
... and also client tests

<azaroth> (And being based on LDP currently means we can inherit many of their existing tests)

shepazu: so we have a variety of tests
... cohesive way of testing would be beneficial

Chris: I had a look at the LDP testing tools
... I would like to see how that fits our testing plan

plh: web platform testing is very active

shepazu: we have a working test suite for LDP, to be re-used
... however, using the tools on webplatformtesting would be more robust

<fjh> +1 to doug

<fjh> +1 to contacting Gregg Kellogg

shepazu: so, should we extend the LDP tests, or should we use the webplatform testing framework?
... the latter would mean converting from JAVA to Python

<fjh> ACTION: azaroth to discuss testing and use of Java vs Python on LDP call [recorded in http://www.w3.org/2015/06/24-annotation-minutes.html#action01]

<trackbot> Created ACTION-21 - Discuss testing and use of java vs python on ldp call [on Robert Sanderson - due 2015-07-01].

azaroth: I will discuss with LDP why they did the JAVA/own testing approach

<fjh> +1 to asking LDP about doing/helping with python appraoch

PaoloCiccarese: maybe they are interested in helping making the change from JAVA to Python?

plh: a lot of the tooling is at most 1,5 year old
... LDP might starting making their tools before
... problem with making your own tests is that they tend to be forgotten

shepazu: [about the infrastructure] you can do proxies, bugtesting, security
... they have a client publishing to multiple services
... it can do a lot

fjh: [talking about when to talk to Gregg]

ivan: Gregg's tools are in Ruby
... it's a whole environment
... what kind of tests would we do?

<azaroth> +1 to coming up with a set of types of test first

<fjh> +1 to having proposal before asking Gregg, good to know Ivan has expertise here

ivan: the same holds for talking to LDP: we need an idea of what kind of tests are needed

shepazu: the question was more: what is the testing methodology for ontology tests?

ivan: Gregg's tools do not handle ontology tests

<fjh> we had a misunderstanding here, good to be discussing this now

<fjh> plh, much thanks for joining our call and giving a useful overview

shepazu: so we should talk to someone with expertise with ontology/datamodel testing

azaroth: Anna Gerber put up a validator to validate the Open Annotation data model
... based on rdf'ing the data model, and run SPARQL queries
... maybe we can transform that into python?

PaoloCiccarese: I also used Anna's queries, easily to be integrated into any language with a lib that supports SPARQL
... but what about the RDF-shapes group?

Protocol

<fjh> CfC to publish FPWD concludes tomorrow, https://lists.w3.org/Archives/Public/public-annotation/2015Jun/0133.html

fjh: we had a CfC for a FPWD

fjh: about protocol, no concerns yet, support noted. If no showstoppers CfC will succeed and we will publish FPWD
... will you be able to prepare the publication draft and arrange publication, doug??

shepazu: yes

<fjh> thanks Doug!

<fjh> http://w3c.github.io/web-annotation/protocol/wd/

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

<KevinMarks> re tests - microformats have a test suite and platfrom hat tests parsers in multiple languages

azaroth: [about the protocol draft issues]
... going through the github issues...

<KevinMarks> https://github.com/microformats/tests

<azaroth> https://github.com/w3c/web-annotation/issues/39

azaroth: #39: right profile in content type for all answers

<KevinMarks> with a runner that calls parsers in multiple languages http://testrunner-47055.onmodulus.net/

azaroth: in JSON-LD, you can add a profile parameter
... that profile can be registered

<fjh> +1 to SHOULD as noted in issue comment thread

azaroth: thought was: if we can give a particular profile, that should be recommended
... so, a SHOULD in the protocol is proposed
... [no objections from the attendees]
... I will close the issues once the edits are made

<fjh> proposed RESOLUTION: resolve issue 39 with SHOULD per comments

RESOLUTION: resolve issue 39 with SHOULD per comments

shepazu: so we can go ahead when the CfC ends
... if no edits are made this week, first publishing could be done next Tuesday

<fjh> note that changes discussed today will not be in FPWD but subsequent WD

azaroth: next issue: about adding changes vs LDP

<azaroth> https://github.com/w3c/web-annotation/issues/38

azaroth: #38
... +1s in the comments

<shepazu> +1

<fjh> +1

azaroth: any disagreement/comments?

shepazu: chris brought up an important aspect: we only need to test the changes vs LDP
... it would be nice if we could structure the test suite as: this is the test suite for LDP, these are the extra tests for the Annotation protocol

<fjh> +1 to limit testing for what does beyond LDP testing, organizing document correspondingly

<azaroth> https://github.com/w3c/web-annotation/issues/37

shepazu: so normative statements about the changes of LDP vs annotation protocol

azaroth: #37 [about supporting more than basic containers]
... cannot be done at the moment
... question is: do we want to allow more, and if so, how to be coherent about 'this is the default, for other requirements, use different containers'
... spec could be more difficult to follow

<fjh> RESOLUTION: Add summary of normative requirements to protocol specification, clarifying those beyond LDP, to resolve issue 38

fjh: so about #38: we add normative requirements about changes from LDP?

azaroth: [about #37] POST message is the same for different containers
... the link headers define what kind of container is returned

ivan: what happens is that you get as response a similar structure as for a basic container, + optional extra information
... client needs to do additional things if it wants to use that extra information
... I suggest to postpone this issue, and talk to experts

azaroth: we use LDP internally all the time

<Zakim> fjh, you wanted to ask why containter type matters

azaroth: I can write something up [after FPWD]

Social Web

shepazu: Social Web WG are interested in our protocol spec

<fjh> https://lists.w3.org/Archives/Public/public-annotation/2015Jun/0231.html

shepazu: activity streams is much more similar to our data model than to being a transport layer
... we should write something about the connection between activity streams and our data model
... maybe activity streams can be used as serialization of annotation data model
... James Snell is interested in working together on this, as well as looking into the annotation protocol spec

azaroth: let's keep discussions centered around the list

shepazu: you can ping him as well

<Zakim> fjh, you wanted to ask about attendees

adjourn