<dape> scribe: dape
McCool: Minutes approval
no objections --> minutes approved
McCool: see https://github.com/w3c/wot-discovery/pull/38
Farshid: Terminology
... e.g., Anonymous TD
<kaz> PR preview - 3. Terminology
McCool: Should order them alphabetically
Farshid: Section 8.2. Directory
API
... added embedded TD
... see Example 1
<kaz> PR preview - 8.2 Directory
McCool: Issue about example/code
highlighting
... normative parts should be visible as is
Farshid: class "advisement" as in wot-thing-dscription/index.template.html ?
McCool: maybe even new class like
"normativeTD"
... see Issue #41
Kaz: Can do CSS updates but should make clear what it means within the text outside the example
<kaz> new issue 41 - Update CSS to add syntax highlighting for normative TD content
McCool: should also look into other specs how it is done there
Farshid: On top of example 1 added
https
... link to anchor possible?
McCool: Should be...
... architecture contains most of our terminology
... maybe also being self-contained makes sense... and not
being vague
Farshid: concern i had, webhooks and long polling is not always working.. for example for sleeping devices
McCool: Yes, notifications might
have this issue
... a scheme like MQTT would make more sense
... or support a scheme where the device explicitly asks for
updates
Farshid: what about websocket?
McCool: Would need to create a
protocol over Websockets
... does not really solve the problem
<kaz> issue 42 - Decide how to securely do notifications from a Directory
Farshid: could provide information
about the time frame (e.g., since yesterday)
... added @type "WoT-Discovery" ?
McCool: Yes, should be part of main ontology
Farshid: Moreover, should it be "WoT-Discovery" or just "Discovery" since there is "Thing"
McCool: see https://github.com/w3c/wot-discovery/issues/43
<kaz> issue 43 - Define the WoT-Directory type somewhere (eg in new version of context, in 1.1)
McCool: context file would be
backward compatible ...
... "Discovery" might make sense
... suggest changing "WoT-Directory" to "Directory"
Farshid: need to take care, have
mentioning before
... will update accordingly
David: in Connexus we used SSE
(works through firewalls)
... sometimes websockets of use also
... we recently enabled websocket
... looks like an SSE stream
Koster: make it look like SSE useful in subprotocols?
David: I think so
<kaz> revisits issue 42 - Decide how to securely do notifications from a Directory
Koster: We encourage people to use SSE also
David: Note: AWS does not support
SSE in lambdas
... do support websockets
... that's why we modified our "convention"
McCool: suggest creating issue in
binding templates w.t.r
... add SSE-over-websockets as subprotocol
<kaz> wot-binding-templates issue 102 - Add SSE-over-websockets as a subprotocol
David: w.r.t. unique ID in
events
... SSE has a field for events ids... but not available for
websockets
... need to add explicit discriminator
Koster: Qualifies it as subprotocol
Farshid: I suggest implementing SSE
first for notifications
... back to Example 1
McCool: Scopes?
Farshid: "write" for changes
... "read" for no changes
McCool: ok, let's revisit this part
Farshid: have action for changing
database
... also read/retriev database
McCool: for htv namespaces one needs to add according context
Farshid: array of responses?
... issue on TD side
... "updateTD" is a simple interaction with the need to know
the "id"
McCool: Anonymous TDs ?
... need to check whether identical
Farshid: Related issue (use-case) exists
McCool: Concern that database blows up if we don't check for identical TDs
Farshid: lifetime is another option
McCool: might have @type for system generated IDs
Farshid: response, fieldValue could be used
McCool: full/relative URL could be
in the response
... should talk about handling IDs in TD
Farshid: we do have full or partial update
McCool: scope "update" besides "write"
Farshid: Yes, makes sense
McCool: what about PATCH?
... op field?
Farshid: op field is optional
McCool: do we have op code for PATCH
Koster: believe we do not have
it
... TD samples are most of the times very simple... no complex
structure
McCool: should also cross reference the idea of using array of responses
<FarshidT> array for response: https://github.com/w3c/wot-thing-description/issues/617
McCool: e.g., error responses
<kaz> issue 44 - Define error responses
<inserted> scribenick: kaz
McCool: editor's note on the issue here?
Farshid: will do
McCool: error responses description here
PR preview - 8.2.2.1 Registration
<Koster> Can we make an Action for the patch case?
[error objects to be JSON-LD?]
McCool: shows TD 5.3.3.8 OAUth2SecurityScheme
McCool: btw, maybe you could
highlight part of the text
... please look at how it's done within the TD spec
Farshid: ok
... should we merge this PR 38?
McCool: few small changes to be
added
... you fix them, and I can merge PR 38 later
... send an email when it's done
... I'll go ahead and merge PR 38
... thanks a lot, Farshid!
... this is a great PR
[adjourned]