WoT-WG - TD-TF -- 09 Sep 2020

<kaz> scribenick: mjk_

Agenda review

Agenda

Minutes from last week

<inserted> Sep-2

Sebastian: status of the rendering script
... other discussion items from last week oneOf in schema, unobserve

McCool: correction to the minutes - "JSON schema *can* validate oneOf

Issue 955

Daniel: there are some limitations in general but JSON schema can check the oneOf constraint

McCool: typo correction

TPAC

Sebastian: start to collect TD topics for TPAC
... JSON-LD specific topics, collect the open questions
... labeling github issues

<kaz> JSON-LD specific issues

Sebastian: versioning topic, how are we going to indicate the version in the TD
... is the context sufficient?

McCool: there are some discovery issues for JSON-LD we could combine

<sebastian> collecting all wot topics for JSON-LD joint meeting: https://github.com/w3c/wot/issues/933

Sebastian: are there other topics for other groups? Security issues with DID?
... what are some deep dive topics for the F2F sessions or topics to discuss in the larger group?

Daniel: we should communicate and discuss the current thinking on dynamic TD/Action status

Sebastian: thing model discussion (TD template)

GitHub issues

<sebastian> https://github.com/w3c/wot-thing-description/issues/958

#958 component mechanism for mashups

Sebastian: why not use link container

McCool: we should parallel the OneDM concept of "Object" (sdfObject)
... also there is a potential for name conflicts with repeated elements of the same type in a TD
... links are one option, another option is a nested structure

Sebastian: it could be related to the thing model topic

McCool: composability of TDs

Ege: related to the research we are doing in putting TD parts into another TD
... also has mashup logic
... can present next week, it is published but maybe links aren't out yet

McCool: there is a youtube
... use cases include modularity, mashup, and discovery by function

<McCool> https://www.youtube.com/watch?v=FGPbl8Occjw

<McCool> sorry, that was on WADE

<Ege_> https://www.youtube.com/watch?time_continue=1&v=yOfA24lbGj0&feature=emb_logo

Koster: describes smartthings use cases
... can share the JSON file for device profile in SmartThings
... common pattern for oneDM and iotschema

<Ege_> https://github.com/tum-esi/wot-system-description

Issue #956 - iri-reference

<inserted> Issue 956

Daniel: it's complex because JSON schema is in flux

Ege: anyuri type maps to iri-reference (iri is a localization of a URI)
... the feature is dropped from json schema due to complex
... it's not well supported and not required for validation
... the implementation in the PG doesn't support iri-reference
... the format keyword is removed and a new mechanism is being created for typing

Cristiano: can remove format and use regex
... but it makes the solution difficult and not recommended

Ege: there is a new $vocabulary term in JSON schema

McCool: suggest to keep things as they are for TD 1.1 and look at new features in 2.0

Issue 957

<sebastian> next issue: https://github.com/w3c/wot-thing-description/issues/957

<McCool> (sorry, I have to drop, have a conflicting meeting I have to switch to)

Sebastian: some protocols don't need unsubscribe/unobserve

Ege: if there is no unsubscribe op, there should be a scripting error
... there could be a default mechanism

Sebastian: how woud longpoll work?

Koster: there are many protocols that share streams, how does a client filter get configured, for example?

Ege: there could be defaults

Cristiano: should be both subscribe and unsubscribe

<sebastian> wow

Ege: we should look algorithmic-ly that I look for the instructions in a form to do what I want to do, before invoking the actual mapped protocol

Daniel: based on this example, how do we describe different locations for subscribe vs. un-

Koster: this would require multiple forms

<kaz> Example 35

Koster: it's not a default but a requirement to always specify some form for unsubscribe, even if it's the same form as subscribe
... this could use an array for the op value "op": ["observe", "unobserve"]

Sebastian: suggest language that each operation is required to have an entry in a form

Cristiano: agree

<cris> server sent event implementation in node-wot: https://github.com/eclipse/thingweb.node-wot/blob/591724445065c7f05b4defb24ebb2a61e3bfd836/packages/binding-http/src/subscription-protocols.ts#L71

<sebastian> next issue: