Meeting minutes
Minutes
<kaz> Mar-7 minutes
Ege: Anybody has remarks on them?
Ege: if none we shall approve the minutes
<none>
<kaz> approved
CoAP
<Ege> PR 352 - Min Polling Interval in CoAP
Ege: The patch was worked on asyncronously
Ege: It showcases we are lacking tooling-wise, we could had the patch completed sooner
Registry
Registry Analysis and Requirements
<Ege> wot PR 1183 - Registry Requirements Revision
Ege: We worked on the document before and we can continue today, but I would like to make sure we agree on the last pending points
<Ege> A binding should be identifiable by the elements in a form such as href, contentType, or other terms. We need to clarify whether the URI scheme and the media type MUST be registered in IANA first. Provisional registration could reduce the overhead. Any new conflicting ones would bring up a discussion but it can still result in "our" provisional getting demoted. Pros: more stable. Cons: More overhead and work A binding that uses a [CUT]
Kaz: You can mention explicitly the patch being discussed
Ege: Yes, the patch is in the wot repository since the work can be shared with other TF
<Ege> rendered text of "Registry Mechanism Analysis" on PR 1183
Ege: we have some todo items before those points, but I'd first tackle the last 3 points
<kaz>
TODO: Clarify what happens when two ecosystems like OCF and LwM2M both
use CoAP binding. The initial thinking is to register them as separate
entries but clarify what they use from CoAP binding. A layered registry
can be thought of as language tags with en extending to en-us and en-uk
where the tags and entries are different but the association to en is
expressed in the id.
McCool: Regarding OCF and its relationship with CoAP, it extends CoAP in specific ways
Kaz: We already started discussing possible extension mechanisms
… do we want to start the detailed discussion now or complete the the global document and then fine tune?
Ege: I'd rather have the big picture look complete and then work on the details later
McCool: An additional question is what to do for protocols using lower level protocols, do they fit a binding or a profile definition?
Kaz: for today, we record our questions as "TODO" and merge this PR itself, then think about how to soleve "TODO"s next. We can identify all the "TODO"s by specifying numbers like "TODO-1" and "TODO-2".
Luca: I'd complete the 3 points first and then use the remaining time for the binding-profile relationship
Ege: We need a way to identify a binding in the TD
Ege: we can use some key-terms such as href, contentType
Ege: It overlaps with the requirement of not having duplicate bindings
Luca: for the bindings we have some terms for the protocol bindings (href and subprotocol) and they may be enough, but for payload bindings contentType is not, since we may want to describe a serialization that in the end is still contentType json
… we should probably use a single term `protocol` to avoid the confusion between the href schema and the subprotocol term and add a term for payload as well in TD 2.0
Koster: I want to add regarding content format: OCF is another example
McCool: We can distinguish the protocol using the url/scheme
Luca: We cannot, e.g. http is ambiguous
<Ege> w3c/
Kaz: We should use "Binding Template" as the title and technical term and think about protocol, formats and ecosystems as features related to the general mechanism
Ege: Let's agree to constrain the bindings on terms contained by Form or Connection
Ege: Next point: A protocol binding submission MUST provide a mapping between its operations and the WoT operations
Luca: I think we can point to HTTP and MQTT as examples on explicit and implicit mapping
Kaz: I was wondering what we want to mean with WoT operations
Luca: Me and Ege were thinking about just the `op` term, you are thinking about the overall operation concept
Ege: Then next point is about formats, their serialization and DataSchema.
… We should make sure that a binding expressing a serialization format provides a DataSchema mapping
Luca: This feels an expansion on the first of the 3 points, we need to add a vocabulary term to avoid clashes on the contentType
Ege: With this we are at the end of the list and I would commit it
Ege: Or given the time constraint we can go over the whole document the next time
<McCool> (ntd, ttyl).
Registries Breakkout
<kaz> Registries Breakout minutes
Ege: We had discussion on existing registries within W3C like the TTWG's registry. We can look at the ttwg boilerplate template
Kaz: I started an initial discussion with plh about that
<Ege> w3c/
Ege: We should also discuss about the binding template section in the TD
<kaz> [adjourned]