03 Apr 2020



Kaz_Ashimura, Daniel_Peintner, Sebastian_Kaebisch, Taki_Kamiya, Michael_Koster, Tomoaki_Mizushima, Zoltan_Kis, Klaus_Hartke


<kaz> scribenick: dape

Past minutes

<kaz> Mar-27 minutes

Sebastian: review of minutes and virtual F2F minutes
... rec status
... issue w.r.t. eventing... templates/common patterns
... talk about efficient formats
... objections?
... no -> approved

REC transition

<inserted> draft for REC

Sebastian: Approval for TD, publish April 9

Kaz: got response from Jeff about quote... waiting for response of Coralie

Sebastian: Perfect
... waiting for some testimonals... having testimonals later is also fine

Kaz: will check the TD document again.. changing the dates et cetera during weekend
... are we OK with CSS settings about highlighting

Sebastian: Yes, look better now
... still some minor issues when it is not valid JSON. Anyhow, it is ok

Kaz: fixed JS and it works

<Zakim> dape, you wanted to with default dynamic

Sebastian: TD draft looks good!


Sebastian: Event patterns, https://github.com/w3c/wot-thing-description/issues/887
... coming from scripting
... common patterns for eventing, eg., subscription ID
... cancelling requires subscription ID again
... Question: can we generalize such approaches
... M. Lagally provided a more complex eventing mechanism
... checked standards such as WebSub
... no ID handling used as it seems
... Yes, likely they keep open socket

Daniel: I believe SSE does not need it either.. keep open connection

Koster: Yes, SSE keeps open HTTP connection with keep alive
... metadata in the stream
... syntactical convention
... could build subprotocol for it
... SSE, WebSub need a sub protocol
... MQTT as well

Sebastian: For MQTT we would not need handler, based on topic
... "data" container might be sufficient
... no subscription/cancellation
... in binding templates we propose the retain flag.. means possible to write property
... for eventing, don't use retain flag simply define event
... kind of guideline
... What about observe CoAP

Koster: CoAP has separate address
... couple of options
... read with observe flag
... believe we have related issues in binding

Zoltan: first sample of event coming right away
... read with event flag

Koster: Unsure what the first value means.. no event took place
... ignore first value

Sebastian: Question is what is handled by the protocol
... e.g., CoAP observe. Binding knows what to do by skipping first value
... this information should not be in TD

Koster: Believe we have related issue

Sebastian: https://github.com/w3c/wot-thing-description/issues/878?

Koster: Initial connection should not be described in TD?

Sebastian: MQTT needs connection to broker before doing anything
... no information in TD how the setup works? Relates to specific binding
... href may contain broker address and topic

Koster: "unsubscribe" should be added in example 34 (binding) or remove op at all
... first time encountering href you could connect...
... cache connection
... security information are still required

Sebastian: Yes, in securityDefinitions ...e.g. basic
... TD gives you the information about security etc. MQTT binding takes this information and setups the actual connection

Koster: Security definition needs broker address also
... could be 3 different brokers

Zoltan: MQTT and OCF could use event interaction and hide the internals

Sebastian: securityDefinitions might contain multiple security information

Koster: Forgot that security can be present locally in form also

Sebastian: Tradeoff between how much information in TD and how much information in binding

Koster: A bit complex but works (caching connections with various security schemes et cetera)

Sebastian: Ben mentioned Mozilla uses own websocket protocol
... essentially don't need form anymore

Koster: Why don't we do that for MQTT as well?

Sebastian: We could
... topic is still needed

Ege: initial connection is implicit in HTTP if you keep sockets alive

Koster: Yes, but there is a header for that

Sebastian: Coming back to original question. We still need a good overview about existing eventing patterns
... looking at Oracle cloud
... CoAP observe
... MQTT
... Issue, https://github.com/w3c/wot-thing-description/issues/888

Ege: Background: property is readable and writable
... read is not value you wrote.. some wrote in between
... in case of oberserve, should I get the value I wrote

Koster: notifications lost?

Zoltan: transaction support

Koster: 1. should there be a notification... I think yes

Ege: Any write request should generate notification

Koster: CoAP creates notification for every update
... there will be open questions w.r.t. protocol

Ege: I just believe we should have some information in behavior assertions

Koster: "Looser" approach works better.. not making too strict assumptions

Sebastian: I am wondering about the relevance?

Ege: testing and behavior. Section 8 of TD also

Binding Template, Ontology for CoAP

Sebastian: https://github.com/w3c/wot-binding-templates/issues/93
... similar to HTTP start work on CoAP

<mjk> https://www.w3.org/TR/HTTP-in-RDF10/

Sebastian: Kaz, can we create a own github repo for ontology on CoAP?
... also who is willing to support this task (so far Klaus and Sebastian)

Ege: Willing to support

Koster: would like to be involved in MQTT also

Ege: Need also someone from MQTT area

Sebastian: at building IOT conference I meet some MQTT people. will try to reach out

Koster: could also reach out to Eclipse founation guys

Kaz: w.r.t. ontology repo: I am wondering about usage and content

<kaz> ontology file

Kaz: move the html file to new repo?

Sebastian: Vocabulary could be used by TD but should not be limited to TD

Kaz: HTTP RDF work kind of abandoned
... do we want to create such work/notes

Sebastian: Would keep work in TD call... having status reports
... charter says 2.12 "protocol vocabulary and bindings"

Kaz: Note: new process within W3C, evergreen process
... let's discuss it in main call

Sebastian: OK, let's do so...

Koster: Could start in binding template document

Kaz: btw, to be strict, "ontology" is different from "vocabulary"

Koster: For our purpose it is equivalent


<sebastian> https://www.w3.org/WoT/IG/wiki/File:2020_TD_Topics_vf2f.pdf

Sebastian: slide #18
... have OAuth2 support in TD but had to remove at-risk features
... "flow" term
... based on GovTech Singapore discussion these missing terms are needed
... idea is to bring it back to the status we had before removing the at-risk features
... node-wot will be offering OAuth2 support also very soon. Work has been started.
... after TD document being published we should bring back this work

Binding templates

Ege: https://github.com/w3c/wot-binding-templates/issues/94
... little problems pointed out
... issue with XML double/number

Daniel: comes from EXI4JSON, can look this up

Taki: will take a look

Ege: https://github.com/w3c/wot-binding-templates/issues/21
... fixed some attribute names

<Ege> https://pr-preview.s3.amazonaws.com/w3c/wot-binding-templates/pull/95.html

<Ege> https://github.com/w3c/wot-binding-templates/pull/95

Ege: ML pointed to term "value scaling"
... not sure about this whole section
... removed 4.4
... also specified OCF version 1.3
... added references (e.g., SenML)

Koster: Could also point to latest OCF version

<Ege> https://github.com/w3c/wot-binding-templates/issues/94

Bank holiday next Friday? Good Friday

Sebastian: Holiday in Europe...
... suggest cancelling the call
... next meeting in 2 weeks

<kaz> [adjourned]

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version 1.154 (CVS log)
$Date: 2020/04/13 07:33:05 $