W3C

WoT Scripting API

18 January 2021

Attendees

Present
Cristiano_Aguzzi, Daniel_Peintner, Kaz_Ashimura, Tomoaki_Mizushima, Zoltan_Kis
Regrets
-
Chair
Daniel
Scribe
zkis

Meeting minutes

Prev minutes

<kaz> Jan-11

Cristiano: about the architecture terminology, we refer to Thing fragments and others about partial TD
... fragment and partial TD are defined in Architecture

Daniel: we need to clarify this

minutes accepted

PR 288 by Kaz

<kaz> PR288

Daniel: PR 288 merged

<dape> closes Isue 284 and Issue 278

Partial TD vs Thing fragment

Daniel: there was discussion on the Architecture meeting

Daniel: a partial TD is similar to a TD, the structure is the same
… a fragment can be anything
… we can pass just a property, without TD context

Cristiano: you got it right

Cristiano: let's see the proposal for the definition in the Arch doc

<cris> https://github.com/w3c/wot-architecture/issues/453#issuecomment-760290680

Daniel: for me partial TD and Thing Model is the same thing, or too close
… a model is like a TD but don't know the instance information
… the context of a Thing model might be different
… the TD TF want a dedicated context for it
… checking validity for partial TD and fragment are simple, just remove parts from the schema

<dape> Comments added to WoT Architecture Issue 453

Cristiano: about the Thing model, I agree. Eventually it could have more features than partial TD
… partial TD is just a runtime concept
… Thing Model can be shared

Zotan: do we mandate the context in Scripting API? currently we don't require it

Daniel: that is correct

Zotan: currently it is a fragment, not a partial TD. Maybe it should be a partial TD

Daniel: without the context it doesn't work

Zotan: we have to update the algorithms then

Daniel: yes, we need to update the wording

Zotan: this validation should be in the TD spec, since it is normative (Scripting is optional)

Daniel: yes, we should reuse a schema defined in TD

Daniel: or in Architecture

Cristiano: I am afraid if we use the Thing Model definition, then we might have some mandatory elements we don't need or want
… it might also evolve with time

Zotan: is context in partial TD?

Daniel: no, I meant the same structure

Zotan: I prefer having an exact schema

Cristiano: yes

Daniel: agree
… we remove the "required" stances and then we are fine

Zotan: will that be the partial TD?

Daniel: yes, if we define it that way in Architecture

Zotan: this is a generic issue for also non-scripting clients, so it needs to be solved in TD

Daniel: not sure they will want to take it

Daniel: the TD spec does not use partial TD
… Discovery, Scripting might use it
… Thing Model is similar

Cristiano: I agree
… we should add the algorithm in Scripting and ping the TD TF if they want to integrate it

Daniel: makes sense

Zotan: OK, we can do it that way

Daniel: we should use the name "partial TD"

Cristiano: I agree with the algorithm approach, not duplicating the schema

Daniel: so how to go forward

Cristiano: I can prepare a PR

Zotan: so we need a section with an algorithm to produce the schema for validation

Cristiano: we can also define PartialTD

Zotan: I think the intent is more generic, for a ThingDescription, it's easier to grasp

Daniel: we can make clear in the algorithm that it is a partial TD

Zotan: we can also use a generic object and describe the algorithm

Cristiano: will think about it

Zotan: do we have a tracking issue?

<dape> https://github.com/w3c/wot-scripting-api/issues/287

Daniel: we used to, changing the title :)

versioning

<dape> ... https://github.com/w3c/wot-scripting-api/issues/224

DP summarizes the issue

Cristiano: I tried to find a use case for versioning, and it's mainly for feature detection
… this would be the main use case
… I agree that modern APIs use feature detection instead
… gave an example in the issue comment
… there might be some complex features that cannot be detected
… and that might need something like a version
… otherwise we should not define a version since we don't have a real use case yet, just hypotethical ones
… also, features should be easy to detect

Daniel: question about feature detection
… any way to test if the server supports something?
… looks like server side feature detection is not needed

Cristiano: right, we don't even know if the server is a WoT runtime at all

Daniel: a second question: how much of a complication will it be

Zoltan: the way we handle values now, I see it as a stable Web platform pattern

Cristiano: right, this is more or less standardized

Daniel: we need more experience and we can discuss later

Zotan: we can close the issue later

Daniel: thank you

adjourned

Minutes manually created (not a transcript), formatted by scribe.perl version 127 (Wed Dec 30 17:39:58 2020 UTC).