W3C

– DRAFT –
WoT-WG - TD-TF

13 April 2022

Attendees

Present
Cristiano_Aguzzi, Daniel_Peintner, Ege_Korkan, Jan_Romann, Kaz_Ashimura, Klaus_Hartke, Michael_Koster, Michael_McCool, Sebastian_Kaebisch, Tomoaki_Mizushima
Regrets
-
Chair
Sebastian
Scribe
cris, dape

Meeting minutes

<McCool> (joining, but working on something out, but yell out to me if you need my input on something)

Previous minutes

<kaz> Apr-6

<Sebastian walks over previous minutes>

Sebastian: any objections?

-> none -> minutes approved

Update TD CR roadmap

PR 1021 - Update TD CR dates

Sebastian: Planned release dates
… agreed to have CR candidate May 6
… CR transition mid May
… that's the new goal

Sebastian: any more comments?
… none -> propose merging PR

Kaz: no objection to PR
… however should try to keep schedule and revisit updates

Sebastian: will do

<sebastian> https://github.com/w3c/wot/blob/main/charters/wg-2021-extension-plan.md

Check status of wide review

Accessibility

https://github.com/w3c/a11y-request/issues/21

Sebastian: no news
… Kaz, can you contact W3C staff ?

Kaz: ok

Internationalization

https://github.com/w3c/i18n-request/issues/171

Sebastian: ongoing

TAG

https://github.com/w3ctag/design-reviews/issues/715

Sebastian: active, waiting for response
… hope to get feedback during CR phase

IANA review for Thing Model content type registration

https://github.com/w3c/wot-thing-description/issues/931

Sebastian: plan to setup all information for TD specification document
… will include all required information for TM
… so far no response
… seems a good sign

PRs

https://github.com/w3c/wot-thing-description/pulls

Default text direction example and assertion #1437

https://github.com/w3c/wot-thing-description/pull/1437

Sebastian: checked before meeting?
… we are still waiting for feeback

Ege: Yes
… suggest to invite aphillips again

Sebastian: Okay. Let's not merge then..

adds hreflang attribute to links #1449

PR 1449 - adds hreflang attribute to links

Sebastian: Got approval already from aphillips
… new term in link container
… e.g., w.r.t. PDF document we can provide language tags
… we also updated an example
… discussed also multiple languages
… in an array form
… not possible at the moment. But new drafts in the pipeline
… aphillips pointed to this new work
… we have an editor-note about this ongoing work in the TD document

Sebastian: I noticed a minor bug with capitalization hrefLang vs hreflang

Ege: I can fix it

Update Implementation Report - Testfest 2022-03 Results #1451

https://github.com/w3c/wot-thing-description/pull/1451

McCool: ongoing work
… can merge and add new PRs for updates

Sebastian: Suggest to merge

McCool: need to re-run manual assertions

Sebastian: merged

Sebastian: What about ThingModels?

McCool: Not yet done work on TM
… need more data
… we still miss Ditto data
… got PR but need to re-run report
… my tool depends on assertion tester

Ege: will talk with Fady

McCool: Should look into CVS files..

fix: allow uri value only for in field of APIKeySecurityScheme #1452

PR 1452 - fix: allow uri value only for in field of APIKeySecurityScheme

Jan: rebased PR
… but I think it is not ready yet
… encountered some issues
… with whitespaces and default values
… I need to figure out the problem

Sebastian: auto has been removed also

Jan: Correct

Jan: Will look into it again

Sebastian: Can look into this PR next week again

Jan: Okay, fine

McCool: need someone to understand the ontology

Security Consideration Update (bugfix, add ref) #1454

PR 1454 - Security Consideration Update (bugfix, add ref)

McCool: minor fixes
… references etc

Sebastian: Looks good -> merging

Add clarification on null schema #1457

PR 1457 - Add clarification on null schema

Sebastian: PR describes what is "null" and when to use it

Ege: found that this description is duplicated

Sebastian: That is always the case
… td-validation and jsonschema

Ege: Does this mean the documents are out of sync?

Sebastian: Don't think so
… but maybe it's good to re-check that

Sebastian: PR looks good
… any objections?
… none -> merging

adds hreflang attribute to links #1449 (Cont.)

<kaz> PR 1449 - adds hreflang attribute to links

Ege: fixed capitalization

Sebastian: Suggest to merge
… grr, there are conflicts

Jan: Seems to be related to the previous PR from McCool

<SK resolves the conflicts>

Sebastian: no objections -> let's merge

Constraining context value according to new algorithm #1458

PR 1458 - Constraining context value according to new algorithm

Sebastian: scripting update?

Ege: old vs new context
… allow both.. only if old one is first
… changes in JSON schema file
… not a normative change

SK/Ege: removed temporary context

Sebastian: Suggest to merge
… done

replaces 'well-formed language tag' with 'valid language tag' (new) #1465

PR 1465 - replaces 'well-formed language tag' with 'valid language tag' (new)

Sebastian: PR clarifies form of language tag
… assertion is updated also
… points out that we require valid language tag

Sebastian: new text also for validation

Cristiano: question about minimal validation
… without network connection ?
… do we have a list of valid tags

Sebastian: It is more about grammar than a list
… like a pattern

Cristiano: Not sure if we should put it as minimum validation
… useful for human
… not so much useful for devices

DP: not sure if pattern is useful

Sebastian: check can also be done by JSON schema

Cristiano: Yes

Sebastian: not mandatory
… just kind of guideline

Cristiano: Got SKs comment

Sebastian: Yep, more a hint with "can be"

Cristiano: The assertion bothers me
… means MUST

Sebastian: addison asked us to use "valid" instead of "well-formed"

Ege: In JSON schema one can put regular expressions
… I can add this
… not sure if resource constrained consumer should do that

Cristiano: I think doing this validation does not give you a lot
… that is my main concern

Ege: for node-wot the developer is responsible
… that's how I see it

Cristiano: I agree
… but we have an assertion
… node-wot should say not-impl ?

Ege: if all examples are satisfying this .. is it not-impl or pass?
… I think it is valid
… like password in description check
… node-wot cannot check this
… we should make such assertions more clear

McCool: manually testable .. different to manually assertion

Ege: Yes, like in Section 8.1

McCool: In theory testable
… with live instances
… Section 9

<Ege> https://github.com/w3c/wot-testing/blob/main/events/2022.03.Online/TD/manual.csv#L57

<Ege> For all security schemes, any private keys, passwords, or other sensitive information directly providing access should be shared and stored out-of-band and MUST NOT be stored in the TD.

<kaz> Preview - 5.3.3.1 SecurityScheme

Ege: Chapter 5.3.3.1
… difficult to check

McCool: A system that does that.. does not satisfy spec

Cristiano: for node-wot ... developer might put password
… node-wot cannot check

McCool: developer is not supposed to do that
… should re-think some assertion if to vague

Sebastian: coming back to minimal validation
… should we keep it in 6.5.1

McCool: seems kind of specific
… wondering.. whether this is too much detail
… why we do not mention other things?

Sebastian: Correct, and we can use JSON schema check

McCool: "for example" is like a spot check
… we should have a final list of checks
… moving things/check to schema is good also

Sebastian: Remove this minimal check?

McCool: no, we should be explicit
… complete this list ... having a finite list

Ege: do we expect all consumers to do these checks?

McCool: Or under "basic validation"

Ege: I think we should agree on definition of "minimal validation"

McCool: definitely JSON Schema
… could stop there ...

Ege: minimal is JSON
… only some spot checks ... not sure about JSON schema

McCool: Mhh, okay.. JSON schema should be under "basic"

Cristiano: I think at minimal ... from usability
… connecting to a Web thing is minimal validation

Sebastian: we are taking time talking about validation
… but back to the point, I would leave the validation process as it is and defer the discussion to 2.0
… about the issue I would remove the sentence added in the minimal chapter

McCool: section does not contains any assertion
… we can mark as informative
… plus I would remove also the last sentence which is mention about tagging assertions
… having some kind of agreement about terms is important
… We also cited that sections in discovery
… therefore I would not remove it

Ege: about what cristiano said. I think it would be ok thinking from the point of view of a consumer

McCool: in a constraint device the process is to check for particular things

Ege: json schema checks can be implemented statically efficiently

McCool: we might explain that small devices does not do a full validation.

Sebastian: ok back to PR
… I remove the validation part
… everything else was not controversial I would merge this

McCool: remember to remove the final sentence about issue tagging

Sebastian: I'll create an issue about that

Issue 1467 created

Accessibility review (revisited)

<Zakim> kaz, you wanted to mention Michael Cooper has responded to Issue 21 https://github.com/w3c/a11y-request/issues/21 (after concluding the current topic)

kaz: before next item, we received an update from the Accessibility Review.

Sebastian: ok

issues

issue #1407

<kaz> Issue 1407 - Additional composition mechanism for Thing Models

Jan: when working with thing model conversion tools, it is a bit difficult to handle composed thing models
… you must store them somewhere
… the issue wants to discussion an addtional mechanism for tm composition

Koster: I'm also doing the same thing

Jan: my original proposal was to use the tm:submodels attribute to store them
… but it did work out

Koster: how did you model a device that has different devices inside?

McCool: HomeAssistant has this hierarchy
… and it is not easy to map to WoT TDs
… we need more experience on building system

Sebastian: I understand your motivation
… but TMs are for code re-use
… maybe in a closed system
… in your proposal how TMs are handled?
… looks like a TM repository
… serialized in the json structure
… also should we discuss this now or defer for 2.0?

Jan: I would be fine for deferring or putting inside another document

Cristiano: same thing happens when you try to instantiate a TD from a TM that is composed by different part
… is not easy to handle links to particular files

Ege: this is also related to json schema
… $ref
… json schema is very strict about the meaning of $ref
… in particular, $ref is just an id not an actual location
… we might solve this using an algorithm for serialization
… like ajv and other implementations of json schema do

kaz: in short we should defer this proposal to 2.0. We are already during the extended charter period. We need to wrap up and close features. This proposal is potentially related to the proposal of grouping devices by ECHONET as well. So we should rather clarify the use case for Ver. 2.0 specs.

McCool: I agree, we didn't have time to understand the problem
… we need to put more work on this

Sebastian: ok
… deferred, but we'll keep an eye on this

McCool: maybe the IG can publish a note
… about this expiremental feature

Sebastian: next week TD call confirmed

<Ege> https://github.com/w3c/wot-thing-description/pull/1437 has been approved by Addison, btw.

AOB

<kaz> [adjourned]

Minutes manually created (not a transcript), formatted by scribe.perl version 185 (Thu Dec 2 18:51:55 2021 UTC).