Key focus topics for the new Charter

[sebastiankb]

In the past, WoT charter work was quite broad and covered many facets with its own task forces such as Architecture, Discovery, Thing Description/Bindings, Profiles, Use Cases/Requirements, Security, API, etc. This was very important in order not to overlook any important aspects and to establish the WoT paradigm correctly. However, due to this broad consideration, we had to organize many web meetings and often the documents could not be published within the requested time of the charter. Based on the development of WoT and its market adaptation, it is becoming clear that the central key technology will be the WoT Thing Description. In this regard, we should focus on the WoT Thing Description in the new charter, as well as what is important for it, such as the Registry, Bindings, and API. I see Use Cases/Requirements and Security as part of the work.

Going this direction, I see an opportunity here to better bundle our resources and publish documents more quickly if the charter has a narrower scope.

I don't want to speak out against the topics of Profiles or Discovery. If we find a good amount of supporters for the new charter, then we should talk about it. In my opinion, these topics would also make more sense once we have published TD 2.0. This two topics could become relevant for the subsequent charter plans then.

[benfrancis]

I had hoped to discuss this in today's meeting but unfortunately we ran out of time.

Firstly, I agree that the WoT Thing Description specification is the core of the Web of Things and that the WoT Working Group could benefit from a narrower scope, fewer deliverables and fewer meetings.

Speaking from my own implementation experience through Krellian and WebThings - we don't really use Binding Templates in their current form, but we do use Profiles. In order for us to adopt Bindings, they need to work better with Profiles, and I think we have the beginning of a strong consensus about how to achieve that. I don't think it will come as a surprise to you that I support a WoT Profiles 2.0 specification being a deliverable in the next charter, since writing the requirements for that specification is the current focus of the Profiles task force that I am currently moderating. I think it would work best if a new HTTP Binding and a new HTTP-based Profile could be specified in parallel, so that they can inform and reference each other. In my view Profiles are essential for real interoperability on the Web of Things, and the Web of Things is not useful to Krellian without them, so waiting another 2+ years before even starting work on Profiles 2.0 would not meet our business needs.

Normative Deliverables

The normative deliverables I would like to see from the next charter are:

• WoT Thing Description 2.0 (Recommendation)
• WoT Binding Registry (Registry, with individual binding documents, including an HTTP Binding and SSE Binding, as Group Notes)
• WoT Profiles 2.0 (Recommendation and/or Registry, with individual profile specifications, including an HTTP+SSE-based Profile, as Group Notes)

I would also like to float the idea of standardising the Web Thing Protocol (WebSocket sub-protocol) in the WoT Working Group during the next charter period. This specification is currently being incubated in the Web Thing Protocol Community Group, but it states in the Community Group's charter that it is intended to eventually join a standards track at a standards body like the W3C or IETF. After six years of incubation we have recently seen a significant increase in interest and activity (including multiple implementations) around this specification and my expectation is that we will finally have a comprehensive draft covering the full set of WoT operations completed by the end of this month. Assuming the next Working Group charter starts in six months time and lasts around two years, it seems reasonable that the WoT Working Group could take advantage of this momentum and take on the task of pursuing this specification along the W3C Recommendation track during the next charter period.

What I think we could drop

Things I think we could afford to cut from the next charter are:

• WoT Architecture - I think this could be replaced by a Group Note or explainer which provides an informative introduction to the building blocks of the Web of Things and how they fit together, with no normative assertions
• WoT Discovery - Although there are some features like geolocation which would be nice to have in a 1.1 version, I don't see a strong need for a new WoT Discovery specification until WoT Thing Description 2.0 is defined
• WoT Scripting API - I still don't see a huge amount of value in this, but I know others do. I don't see any harm in updating it if people are motivated to do so.
• Security & Privacy Guidelines - Could be integrated into the Privacy & Security Considerations sections of individual specifications

Something else which could help reduce some admin overhead would be to discontinue the WoT Interest Group and split its responsibilities between the WoT Working Group and WoT Community Group as appropriate.

---

Backup Plan

If there isn't sufficient interest (or resources) to work on WoT Profiles in the next charter period, then another option I would consider would be to define an HTTP+SSE sub-protocol as part of the Web Thing Protocol, as an alternative to a Profile. This would allow the WebThings project (and others) to implement the Web of Things without using Bindings or Profiles at all. This has always been within the charter of the Web Thing Protocol Community Group and defined in our Use Cases & Requirements document, but our stated preference has always been to work more closely with the WoT Working Group if possible.

[benfrancis]

@sebastiankb wrote:

we should focus on the WoT Thing Description in the new charter, as well as what is important for it, such as the Registry, Bindings, and API

If the motivation is to focus entirely on Thing Description 2.0 (and its bindings) in the next charter period, then I think if Profiles are removed from the charter then the Scripting API must be removed too. Scripting API is not a normative deliverable, is not essential for producing or consuming Thing Descriptions and is fairly JavaScript-specific so can not be used by all implementations. I don't object to people continuing to work on the Scripting API, but I do think it would be wrong to keep it in the charter in favour of Profiles, which I and others see as vitally important for interoperability on the Web of Things.