W3C

– DRAFT –
WoT Discovery

07 February 2022

Attendees

Present
Andrea_Cimmino, Christine_Perey, Cristiano_Aguzzi, Ege_Korkan, Farshid_Tavakolizadeh, Jan_Romann, Kaz_Ashimura, Kunihiko_Toumura, Michael_McCool, Tomoaki_Mizushima, William_Holroyd
Regrets
Christian
Chair
McCool
Scribe
cris

Meeting minutes

New participant

Kaz: we have William as a new member

William: I'm working on client and cloud software at Lenovo

McCool: thank you for checking with us, here we are working on discovery

McCool: but on Wednesday we have a broader call with all the Task Forces

McCool: we are on a clean up phase

minutes review

<kaz> Jan-31

McCool: we have just one section, it is not usual
… probably we forgot to add subtopics

<mccool pointing to possible sub sections names>
… we discussed about ThingLink and use cases
… remember that we had a resolution for freeze everything except the ThingLink topic
… minutes oks, please fix the sub topics

Pull Requests

Issue 256

<kaz> Issue 256 - Small fixes on Creation

McCool: don't worry about 256 is informative

PR 269

<kaz> PR 269 - Permit use of TD Links for Self-Description of Multiple Endpoints

McCool: basically we have a discovery process composed by an introduction and an exploration phase
… we identified a corner case that we want to cover analyzing the /.well-known introduction mechanism
… in particular, it does not cover having multiple TDs on a single enpoint
… for example an outlet strip which has different Things for each outlet

William: is this something about handling sub-things from a root component?

McCool: I played a bit with it and I think it is better to model does sub components as sub things not sub affordances.
… back to the use case we have a Directory API service but it is not really suited for this small static TDs

McCool: to get a better understanding of the use-case let me review comments
… Ben's concern was that he can only point to a directory was to use a ./well-known endpoint
… adding a thing link might introduce complexity to the client
… we might not put the Discovery API optional
… also link should not contain affordances
… a Thing collection only apply to thing models
… it is flatten in thing models

Farshid: Ben's concern is not about parse the list, but more about the fact the registration api is mandatory
… then also understand the TDs

McCool: I agree the read-only thing should be fixed
… the issue about understanding the TD if the TM is normative a consumer can just cherry pick what it needs.

Farshid: yes, but still it needs to read the base path

McCool: yes but reasonable

<FarshidT> Multiple Responses from Peer-to-Peer Discovery: https://github.com/w3c/wot-discovery/issues/249

Farshid: I can work on the read-only point
… please refer to the issue above

McCool: the title is not perfect
… we should fork to another one
… returning an array from the .well-known/ should cover the issue

Farshid: actually the spec from well known allow that

McCool: the ThingLink was equivalent to the array
… just because other introduction methods return always a TD
… partial thing description directory might be also impractical implementation side
… so we have the option of making the TDD API has everything optional
… but only read operations mandatory

<wholroyd> TD is Thing Directory, correct?

McCool: still need two fetches

<FarshidT> TD: Thing Description

<FarshidT> TDD: Thing Description Directory !!

McCool: but a good nonetheless

:) we are full of similar acronyms

McCool: I would move listing chaper outside
… together with retrieval

Farshid: retrieval might be not so critical
… it might add complexity to the implementation

McCool: point taken
… but listing and retrieval go together
… is the TM requiring something?

Farshid: we have a tm require
… I can do the PR

Cristiano: I'm glad that we are introducing this
… but we are not covering TM model compositions

McCool: I think discovery should just strictly focus on flat lists of TDs
… plus we are missing a formal client discovery described in the spec

McCool: I can work on it
… but we cannot add normative statements

<McCool_> proposal: resolve issue https://github.com/w3c/wot-discovery/issues/272 as well before normative feature freeze

<wholroyd> Reading this spec is like trying to learn English by reading a dictionary

<wholroyd> What is the best way to propose a series of questions to resolve some issues I have

RESOLUTION: resolve issue https://github.com/w3c/wot-discovery/issues/272 as well before normative feature freeze

<wholroyd> ...where issues I have is things I don't understand yet

<McCool_> regarding wholroyd's point, I will also try to fill in some missing informative sections as part of resolving issue #272

<FarshidT> @wholroyd, would be best to search (https://github.com/w3c/wot-discovery/issues) in open/closed issues and file new issues for each remaining question.

<kaz> [adjourned]

Summary of resolutions

  1. resolve issue https://github.com/w3c/wot-discovery/issues/272 as well before normative feature freeze
Minutes manually created (not a transcript), formatted by scribe.perl version 185 (Thu Dec 2 18:51:55 2021 UTC).