Re: Regular Web of Things call on scripting API

On Mon, Nov 7, 2016 at 2:43 PM, Benjamin Francis <bfrancis@mozilla.com>
wrote:

> Do you imagine this API being implemented on the client side (e.g. as a
> DOM API in a browser engine) or on the server side (e.g. as a Node module)?
> If it's the former, has any browser vendor expressed an interest in
> implementing this?
>

The ExposedThing interface is on server (sensors, actuators) side, in order
to facilitate writing support for new Things. We do have an implementation
to a very similar API (OCF) in a constrained JS runtime (based on
JerryScript) that runs even in MCU's.
https://github.com/01org/zephyr.js/



> If it's the latter, what is the benefit of standardising this scripting
> API? Calling a REST API is likely to work differently in different
> languages, of which JavaScript is just one.
>

The client side interface is a ConsumedThing, and could be in fact a
library on top of REST. Anyone can replace it, i.e. no need to standardize,
but it would be nice however.

Then, nobody prevents standardizing scripting for other languages.

We agree that the mother of all is the REST API.


> Rather than specify the scripting API, have you considering specifying the
> REST API in terms of methods, requests and responses instead like the OCF
> does?
>

We do have a REST API server, and I could imagine WoT comes up with a
similar (but higher level) solution.
https://github.com/01org/iot-rest-api-server/

The draft I created is deliberately based on REST model (and connected to
OCF implementation, although not the same).

Note that the links above point to OCF implementations, not WoT
implementations. Nevertheless, we do plan to support WoT scripting, once
there are concrete use cases and a scripting API draft that addresses those
use cases.

We have discussed these in today's Scripting API meeting.

Best regards,
Zoltan