<kaz> scribenick: andrea_cimmino

Prev minutes

<kaz> June-1

meeting starts with reviewing the meeting minutes of June the1st

objections for publishing the minutes of the 1st of June are not presented

following the minutes of the meeting of June the 15th are reviewed

<kaz> June-15

no objections to publish also the minutes of the 15th of June

<kaz> vF2F minutes

finally, the F2F draft minutes are reviewed

McCool: recaps the discussions: requirements and design decisions, LinkSmart directory service, and DID

<McCool> Verifiable Credentials in IoT Services

McCool: requests the change of one subsection named DID for Verifiable Credentials in IoT services

<McCool> https://github.com/w3c/wot/blob/master/PRESENTATIONS/2020-06-online-f2f/2020-06-22-DID-VC-Fotiou.pdf

kaz updated the title and liked to the slides

no objections for publishing the minutes of the F2F meeting

New requirements from F2F and issues

<scribe> new ideas for the requirements extracted from the F2F and issues

There are scenarios in which with a link has a type and thus, beforehand it can be know if the link points to a device or a directory

McCool: do we want to have someway to distinguish this scenarios?
... ...we should add UPnP as an introduction mechanism (extracted from F2F minutes)

<kaz> wot-testing repo

<kaz> kaz: we could look into the wot-test repo issues as well :)

<inserted> mm: any specific topics in your mind, Farshid?

farshid: idea regarding modelling the directory structure

proposed by FarshidT

McCool: adds a new topic of discussion proof chain (from F2F minutes)
... adds as a discussion topic to define the purpose of TTL
... adds as a discussion topic the streaming for delivering large TDs from directories
... adds as a discussion topic "local vs global" IDs
... adds as a discussion topic OpenID

the following requirements capture the requirements from the minutes of F2F

correction*: the previous requirements capture the requirements from the minutes of F2F

from the wot-testing also new requirements/discussion points are derived

McCool: adds the discussion about "IDs from TDs troubles as primary keys since they are optional"
... adds as a discussion topic the Json Path implementation and incompleteness problem
... adds as a discussion topic the DNS-SD service types
... now let's walk through the previous ones to take decisions

<kaz> list of topics:

<kaz> Untyped introductions

<kaz> Links in directories

<kaz> UPnP as introduction

<kaz> Acronym for Thing (Description) Directory; "TD" is no good

<kaz> Proof chaining

<kaz> Define purpose(s) of TTL; protocol-specific requirements

<kaz> Streaming (for delivering large TDs from directories...)

<kaz> Local vs. global IDs

<kaz> OpenID

cris_ idea on topic Untyped introductions, "having the types available"

Regarding Links in directories, likely is better to have links to TDs and no TDs themselves

FarshidT points out that maybe the opposite alternative could be better for scenarios where queries should not be sent to devices

in order to be solved over the TD

McCool: ...the bottom line, does any of this discussion requires to create an issue?
... UPnP as introduction probably requires
... regarding the acronym, "TDD" could be better
... "Thing Description Directory (TDD)"
... Proof chaining likely will require a PR, seems reasonably to be considered
... Regarding Streaming will require also a PR
... regarding OpenID maybe is not a discovery problem, more a security problem

we move to check the issue from Ben

McCool: we will have a discussion in regards the difference of "directory" and "gateway"
... gateway is view as a box in which different services can be run
... the directory is a service, thus, it could be ran over a gateway
... from the points posted by ben the first and second are highly related to discovery, but not the third which could be out of scope

<kaz> Issue 32 - What is a Thing Directory?

McCool: answers the issue explaining the difference between using gateway as a service, or as a device
... adds to the issue answer the definition of what is a directory as endowed by the group and its difference with the gateway
... includes that it would be ok to have only links in the directories, and motivates this statement
... concludes that having only links is not as suitable as having TDs when comes to discovery

McCool is going to post the issue for discussing having a list of links and TDs stored at the directory

kaz suggest to look into Manifests and review the IETF rfcs and drafts

<kaz> McCool's comments for Issue 32

<kaz> [adjourned]