Meeting minutes
Minutes
March 11
Lagally: (updates the agenda wiki)
… starting with the March 11 minutes
Lagally: minutes approved to publish
vF2F minutes
<kaz> vF2F Day 3/4
McCool: will refactor the terminology document soon
Lagally: any other concerns besides the one typo?
McCool: Philippe Coval name seems mis-spelled
McCool: what is Philippe's role? is it listed correctly as a guest?
McCool: make sure he is listed for the day in which he made the comments
<kaz> (added Coval and fixed typos)
Lagally: any objections to approving?
… approved
Issue 73
<kaz> wot-profile Issue 73 - Refine Goals and Scope
Lagally: reviewing the profile goals and scope issue discussion
Lagally: let's keep the discussion focused and avoid CoAP, MQTT< etc.
McCool: agree, let's meet the needs for webthings as a priority
Lagally: is there any issue with the proposal around http, JSON, and default protocol binding
Sebastian: this looks good for a http/JSON baseline and template for creating future profiles
… we should adopt only what really makes sense based on the experience we have
… we need to discuss websockets
McCool: websockets is not yet developed in our group
… no really good agreed solution
… let's focus on the non-controversial points
Sebastian: maybe the items listed in brackets are considered optional
McCool: we could leave events open to implementation
McCool: http could use long polling for observe
… OK if the profile leaves some things open
Lagally: read, write property and invoke action only
… bare minimum
Implementations and test
Lagally: is there any more input from the plugfest?
McCool: there was activity around TM conversion and discovery, not as much about device to device
… June plugfest will be focused on higher level questions also
<kaz> doodle for the next testfest
McCool: to test the profile we will need a test plan and two implementations
McCool: to reduce risks we can keep it simple
McCool: can replicate the current assertion mechanism for discovery and profile
McCool: we should break out the architecture assertions separately
… each spec should have its own plan, report, assertions...
Lagally: what's the process for marking up for the assertions?
McCool: look at the TD one for examples
McCool: tables have some particular patterns
McCool: wait to see what comes from webthings
McCool: the implementation report is generated as a snapshot
… what are the architecture assertions?
… terminology is normative
Lagally: issue #74 can be linked to the issues #75 and #76
McCool: the tooling needs to be set up
… once set up, the tooling can identify assertions by the span markup
<kaz> wot-profile Issue 74 - Highlighting assertions
<kaz> wot-profile Issue 75 - Implementation Report
<kaz> wot-profile Issue 76 - Markup of normative requirements (RFC 2119)
Profile
<kaz> wot-profile Issue 73 - Refine Goals and Scope
McCool: it's OK to publish a small and simple profile spec to start with
Lagally: we need to see what is in the webthings proposal
McCool: the outline looks acceptable
McCool: any objections?
Sebastian: there are more topics to discuss, so we can leave this as is for now
Lagally: we need to define action and error behavior
Sebastian: should we wait for feedback from Ben?
McCool: it could add 2 weeks delay
McCool: let's note our agreement to adopt the structure of "Protocol Binding" in the issues
Koster: agree with the things not in brackets, and the bracketed items can be left to implementations
Koster: also agree that templated URIs and data schemas can be included somehow
McCool: that's a longer discussion
Lagally: also is TD scope
canonical TD
<kaz> wot-thing-description PR 1086 - Add section to define Canonical serialization
McCool: still working on the specification of the canonical form
… identifying assertions and thinking about implementation
… need to find a library to process JCF
McCool: there is an issue with how to handle the defaults in TD extensions, like http vocabulary
… defaults are required to be omitted in the canonical form
… RDF processors will fill in the defaults
… it makes round-trip processing an issue
… also an issue of regenerating prefixes
… (discussion around tradeoffs and issues in TD canonicalization)
Lagally: canonicalization won't impact the profile spec
Lagally: the profile needs to restrict content types
Lagally: should the profile allow communication without TLS?
<kaz> WoT Profile Editor's draft - 5.2 Protocol Binding
McCool: the local hub server can't use TLS because TLS requires some PKI infrastructure
… we could provide object security with raw keys
McCool: browsers require root certificates
McCool: the base plan is to use HTTP without TLS
… we could allow http but highly recommend https
Lagally: is there a way we can require https for remote interactions?
Lagally: we seem to be reaching a common understanding
… time to close the meeting
AOB
Lagally: there are a lot of outstanding issues
<kaz> outstanding issues for wot-profile
McCool: maybe we can defer a lot of these and some may not be relevant in the context of the simple profile
Lagally: any other business?
… adjourn