Meeting minutes
Agenda bashing
https://
Minutes approval
https://
Sebastian: talked about publication plans
… VF2F TD topics
… looked into PRs
… merged some PRs
McCool: some security PRs should be ready by next week (expect review from Oliver)
Sebastian: RDF PR might be closed... need to check with Victor
Sebastian: looked at issues
Sebastian: AOB about Ege / Henry and JSON schema definitions
Ege: had meeting with Henry..
Sebastian: Objections to approve minutes?
-> none -> minutes approved
F2F minutes
<kaz> Day 6
Sebastian: talked about round tripping
… Victor could fix all issues (RDF <-> JSON-TD)
… gave overview about 1.1 features / updates
… talked about thing models
… use-case from Eclipse Ditto
… McCool gave presentation about validations
… talked about action model
… sync vs async
Sebastian: objections to merge?
-> none -> minutes approved
WoT Bindings
Ege: main point is Modbus binding
<Ege> https://
Ege: Cristiano created PR
… binding is reviewed also by Sebastian
… looks good
… talked with Cristiano about schema validation
… concern for other protocols as well
… make sure to use the right terms
… TD is flexible w.t.r.
… Cristiano provided schema file
Cristiano: >shows rendered version<
… improvements from last time is introduction of schema
… we have SHACL
… we have JSON schema file
… made some assumptions
… TD needs to use same prefix
… otherwise we need pre processing step
… JSON schema checks context url and prefix
Sebastian: We could check with Victor about other solutions
… prefix usually can be used randomly
<McCool> (I'm back)
Cristiano: For SHACL file there is a specific place
… for JSON schema this is not the case
Sebastian: wonder about canonicalization and how it is done there
McCool: 1. keep prefix or 2. substitute prefixes
… 3. create new names for all Urls
… other concern I had was with having multiple schemas
… how to do proper validation
Cristiano: JSON schema needs to be replicated
… in form level "own" schema is added
McCool: What about more extensions
… we look for a wrapper schema
Cristiano: Having one schema makes validation simple
… saw other use-cases on the Web doing the same
McCool: using canonical form for validation?
… fills in default values
… validation should be done before canonicalization
… validation process could be some specific pre-processing
Sebastian: Maybe we can talk with JSON schema people
… suggest we keep the current proposal
… keeping prefix "modbus"
… should be made clear in text
McCool: template would make this behavior clear
Cristiano: having {{modbus}} template cannot be used as it is
Ege: did some tests with JSON schema
… having Modbus and non-Modbus seems to fail
… "modv" is also used in some examples
Cristiano: will take a look, thanks
… suggest to merge the PR soon.. and move on from there
Cristiano: added small readme also
Sebastian: support merging
Daniel: noted that github "preview" link does not work any longer
<cris> https://
Cristiano: Yes, would be good if it can handle files other than index.html
Kaz: Let's merge and see whether it can be resolved
… comment: the header should not link directly
Cristiano: temporary situation
Ege: OK with merging also
… no more concerns mentioned
… will do the merge
Sebastian: use same pattern for other bindings also?
Ege: Yes
PRs Overview
Add subscribeallevents and unsubscribeallevents operations - closes #1082
https://
Ben: helps getting WebThings platform to be compliant
… observing in one operation
… a bit unclear what has to be modified manually.. and what is auto-generated
… I think I got it right in the end
Sebastian: looks good to me
Ben: updates for architecture are to be followed
<cris> +1 for merging :)
Sebastian: Any comments?
… looks we can merge
… thank you Ben
correct <span> element for ComboSecurity assertion and add unique id #1184
https://
Sebastian: seems to be a typo fix
Ege: yes, span vs scan
Sebastian: Fady is not yet an official W3C member
… we work on fixing the situation
Kaz: Do we want to wait or continue.. marking as non-substantial
… given this is just a typo fix merging seems ok
Sebastian: merging ...
fix: replace "name" with "title" in TM example #1175
https://
Sebastian: fix is okay
… issue that it is changed in index.html only
Kaz: editorial fix again, but we need to fix the index.template.html as well. so would suggest you create another PR to fix the index.template.html (and index.html as the result of the tooling)
Sebastian: will let Jan know that both files have to be updated
TD Signatures #1151
https://
McCool: Oliver will take a look
… still some things to do
… using "Signature" term caused conflicts
… terminology vs vocabulary
… issue in Section 5.2
… will create issue about that
… once we get feedback from Oliver the rest should be rather minor
Sebastian: will also ping Oliver
Other PRs
Sebastian: other PRs are about auto-updating and such
… changing package versions
McCool: should test these changes before merging
Sebastian: Daniel will have a look
WIP: swap securtiy and securityDefinition in context file
https://
Sebastian: should check with Victor
McCool: Yes, I think this should be fixed by other PRs
… but let's ask Victor
Issues
Describing initial connection #878
https://
Sebastian: Old issue
… metadata about initial connection
… subprotocol et cetera
… WebSocket, MQTT, ... mention the endpoint ... no more special information on interaction level
… idea: using top-level information
… example from Ben by means of WebThing
… no need to repeat the same information over and over again
… we have "base" but we might need more complex structure
… introduce new term "bases" ..
… define endpoints and refer to it on interaction level
… another proposal from Ben .. using global form
… more opinions / solutions
McCool: 2 issues come to my mind
… what about security
… initial security bootstrapping
Sebastian: forms can contain security
McCool: Yes, but we might need to override it
… first getting thing link
… afterwards the full TD
… anyhow, security should be considered for these examples
Ben: i don't think we have security bootstrapping issue
… multiple bases is orthogonal
… this is more about having a persisting connection
… in case of WebSocket and subprotocol it should know when to use it and to setup just one connection
Cristiano: I reviewed the proposal
… none is solving all problems
… moreover, backward compatibility is also an issue
… shift some features to 2.0
… otherwise go on with Bens proposal that does not breaking changes
Sebastian: backward compatibility is a real issue. right
… ignoring is ok... but interoperability is not working
Ege: we have 2 things
… base
… and initial connection
McCool: what about using action for onboarding
… can be used by "older" implementations
Ege: How to know that one actions depends on another action
McCool: needs to be labelled
Ege: node-wot for example would not keep connection
… no good solution for old consumers
McCool: Do we properly support MQTT in 1.0 ?
… if not, no need to further tackle this
Ben: Initial problem: never really defined properly... everyone doing it in its own way
… hence no backwards compatibility
McCool: Agree
Ege: minimal change ... old consumer should not fail
… eg. uses bases and not base
Cristiano: I was thinking more about syntax
… forms at root level seems less intrusive
Ege: does not solve multiple base
Cristiano: Yes, should be post-poned to 2.0
<Ege> https://
Ege: Bug in current editors draft (see link above)?
Ben: what does "with default" mean
Ege: if nothing is defined, default kicks in
Sebastian: I was thinking about "initconnection" op value
Ege: consequence would be to check first about "initconnection"
… mixes concepts
Cristiano: href required by TD .. but not by WebSocket
Ege: Yes, having empty string
Ben: Yes, forms is mandatory.. not needed by WebSocket only
Sebastian: Another path could be saying we need to post-pone till 2.0
… or form needs to be present for now
Ben: my goal is to make WebThings compliant
… option 1: using link
… option 2: use bug with empty forms
Cristiano: should first talk how to handle the bug
McCool: like the suggestion to have a new op (like proposed by Sebastian)
Ben: in the case of WebThing... the op is "all"
… single websocket connection
Ege: kind of a mis-use of semantics
… why not listing them all
<benfrancis> https://
Ben: created issue, see link above
Sebastian: wonder why it closed?
Ben: I did close it.. assuming op can be blank
McCool: "all" is bit weird
… maybe you just want to do some
… still like op "open" or so
Ege: How to know when to do this form
Cristiano: once it is there.. every affordance should use it
McCool: same protocol
… using same connection
Cristiano: WebThing has just top-level form...
McCool: how to distinguish href?
Ben: are the same
… pubsub in WebThings is not yet fully supported
… looking for simplest solution
McCool: having top level form looks like simplest solution
Ben: if we had op "open" and forms in each interaction. Does this work for MQTT?
Ege: makes sense
McCool: does prefix part indicate broker?
Ege: Yes
Ege: annoying problem. mqtt://
… in form in property myThings/temp it is not set whether to use with or without slash
McCool: using fragment identifier may help
Ege: might work
… in MQTT "#" is special... like wildcard
… should use "$" instead
McCool: MQTT never sees the pound sign (#)
Sebastian: Suggest getting back to global connection information
… does new op type work
McCool: Looks like
Daniel: wonder whether we need the counterpart "close" too
McCool: Suggest to wait for use-case
… or call it "initialize" to avoid questions
<McCool> (another option would be "connect" for the global op)
Ben/Ege: >discussions about WebThings and base<
Ben: not sure about the problem we are solving
MK: WebThing still needs selector
McCool: Form is required .. but for example form can be inferred.. also in ECHONET
Cristiano: problem we solve: put form there for WebSocket
Ben: for http and WS we do not have a problem
Sebastian: suggest to continue the discussion in issue
… root form looks promising
<McCool> (sorry, another meeting, need to drop)
<kaz> [adjourned]