W3C

WoT Discovery

07 Dec 2020

Agenda

Attendees

Present
Kaz_Ashimura, Michael_McCool, Farshid_Tavakolizadeh, Jack_Dickinson, Kunihiko_Toumura, Tomoaki_Mizushima, Andrea_Cimmino, Michael_Koster
Regrets
Christian_Glomb
Chair
McCool
Scribe
FarshidT

Contents


<kaz> scribenick: FarshidT

minutes

<McCool> https://www.w3.org/2020/11/30-wot-discovery-minutes.html

no objections to publishing the minutes

scheduling

McCool: on vacation starting next week. Someone else could chair

Farshid: I can chair if we have other things to work on. Let's see how the meeting goes today

PR: Updated document with federation (https://github.com/w3c/wot-discovery/pull/105)

McCool: there are a couple of issues.
... writing comments on the PR...
... suggests breaking the current federation assertion into two assertions.

Andrea: I agree, will change it for the next meeting

McCool: also, use "WoT Thing Description Directory" consistently in the document
... commenting on How do we handle federation in SPARQL queries (https://github.com/w3c/wot-discovery/issues/97)
... linked the PR to close the above-mentioned issue automatically

Issue: Links (https://github.com/w3c/wot-discovery/issues/34)

McCool: looking at https://github.com/w3c/wot-discovery/issues/34#issuecomment-668508701
... One constraint is that we are reserving the type "Link" which prevents others from using it
... we can store the TD of another directory, links are not necessary here

Andrea: it is necessary to have a distinct type for directory TDs

McCool: what is the use of having Links instead of TDs inside the TDD?

Farshid: it could be for privacy reasons, highly dynamic TDs, very large TDs
... the link could point to a Thing or Directory TD

McCool: documenting use cases on the issue
... security and securityDefinitions are mandatory but useless if we only have links. I argued against making it mandatory at the time.

Koster: we could also use URI instead of Link value for type

Andrea: there is already a Link type, so there will be collision

McCool: few cons, what are the pros?

Farshid: all existing mechanisms for TD will work out of the box

Andrea: why do we need a new type for links? why not just use the Thing/Directory types?

McCool: we could look at relation type to know too.
... but it is good to know it in advance

Koster: there could be a way to have alternative TD schemes without mandatory TD fields

McCool: another examples is the storage of Thing Models
... I think @type with expanded URI is the best way to go

Farshid: A full TD will allow adding metadata which is useful for search.

McCool: also for adding metadata (annotations) to other internal TDs without changing those TDs
... comments: https://github.com/w3c/wot-discovery/issues/34#issuecomment-740004060

<McCool> proposal: use a special "Link" TD as discussed in https://github.com/w3c/wot-discovery/issues/34 to refer to TD held elsewhere from a directory

RESOLUTION: use a special "Link" TD as discussed in https://github.com/w3c/wot-discovery/issues/34 to refer to TD held elsewhere from a directory

McCool: suggesting TDLink or TDPointer

Koster: TDPointer is usually for only a reference

McCool: to create a PR and specify the details
... added example for the resolution: https://github.com/w3c/wot-discovery/issues/34#issuecomment-740009713

next meeting

McCool: canceled. Next meeting on January 4th

<kaz> [adjourned]

Summary of Action Items

Summary of Resolutions

  1. use a special "Link" TD as discussed in https://github.com/w3c/wot-discovery/issues/34 to refer to TD held elsewhere from a directory
[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version (CVS log)
$Date: 2021/01/11 04:50:13 $