W3C

WoT Scripting API

19 April 2021

Attendees

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

Meeting minutes

Previous minutes

<kaz> Apr-12

Daniel: last time we reviewed previous minutes from vf2f and regular meeting
… we agreed to plan joint calls, but I still need to talk with each taskforce
… then zoltan proposed to improvements to the discovery api. There's a PR in flight about it.
… any objection to merge the minutes?

Zoltan: they are fine

Daniel: minutes approved

quickly updates

Daniel: other task forces are ready to publish updated versions of their documents
… about the Scripting API document, it seems that there are no sensible updates so far

Zoltan: since we depend to other documents it is normal to be a little bit late
… we can publish later

Daniel: yeah we can wait until other documents are more stable

Cristiano: +1 for postponing

<Mizushima> +1

Mizushima: ok

open PRs

Daniel: zoltan update the discovery api, it is a substantial change
… no more local

Zoltan: added direct

<kaz> PR 316 - Simplify discovery methods. Update examples.

Daniel: it's similar to what we had a while ago (regarding direct)
… direct it's difficult to formalize because there are a lot of parameters that a user can provide (url parameters, headers ...)

Zoltan: right now url is just a filter
… plain string, the runtime should check it to understand the address
… cristiano also proposed to refactor the APIs using the methods instead of filters
… it's another paradigm more from js frameworks
… in the web platform is more usual to use filters

Daniel: direct method could be a lot
… can we support api to understand if we support coap/http direct?

Zoltan: we can't control the binding here

Daniel: in node wot we had a method to fetch method, that cause some problems

Zoltan: in ocf direct method is well-formalized even parameters
… we should support
… it

Daniel: ok, I'll check the PR later and we can discuss the new api design later

Zoltan: I was concerned about having an API that fit the current discovery spec, we refactor later

Zoltan: I also have a PR (closed) for 'once' subscription method. keep it mind and provide feedback if any

Daniel: kinda supported, you can unsubscribe right after the first event

Issues

taskforce related issue

Daniel: I created issues to capture discussion topics for each defined joint call with other taskforces.

<kaz> Issue 313 - ThingDescription TaskForce related issues #313

Zoltan: the discussion list for TD taskforce looks good

<kaz> Issue 314 - Discovery TaskForce related issues

Daniel: about discovery we have discovery methods, streaming TDs, and fetching mechanisms
… fetching is a broad topic

Zoltan: I would rename it to something else

Cristiano: what about just fetch?
… I opened an issue about it in the discovery repo

Daniel: added
… streaming is another big issue

Cristiano: yeah it's an hot topic in the Discovery TF

Daniel: I remember that mccool was discussing about field order when streaming

Zoltan: it seems out-of-topic for us

Daniel: we could allow clients to request for streaming tds

Daniel: another topic, with discovery discovery can select pieces of TDs

Zoltan: yes please add it

Cristiano: yeah, I think it has higher priority than streaming

Zoltan: I would remove the first topic

Daniel: ok

Zoltan: cause those are the real question

Daniel: we could try to join next week discovery call

Zoltan: the earlier the better

Daniel: let's set the meeting for the 26th of April

Zoltan: should we do also the security call?

Daniel: it's ok

Zoltan: it's like a big scripting call 3 hours with one hour of break

Daniel: yeah it's good

Daniel: about the security joint call
… there are two topics managment api and choose the security scheme
… kinda related

Zoltan: that's off topic

Daniel: I know

Zoltan: should we try to draft this management api before the call

Daniel: we could start from the node-wot
… in the discovery they started from a TD
… we could do the same

Zoltan: I have to think about it, I would be cautions

Cristiano: I'm afraid that some managment methods should never be called remotely
… so I am not sure that starting from a TD is the best way forward
… we could have a TD after
… but still we can look to node-wot as a starting point

<kaz> i|I am afraid|Issue 298 - Requirements for Managment APIs|

Zoltan: we have a lot of different deployment methods: docker, web assembly, web manifest etc.
… it's onboarding I described it in the architecture
… basically a device has two mode (one for provisioning and one for runtime)

Cristiano: I would say that providing a draft of the management API it's beneficial

Daniel: I don't think we have time to define it

Zoltan: we don't even have good use cases
… I am more concerned with the discovery.

Cristiano: we could start from security use case that we had for the managment api
… I can "formalize" it

Zoltan: we could start from it, write code and than switch to the runtime mode
… ExposeThing is already at managment level api.

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