<scribe> scribe:zkis

Prev minutes

<kaz> June-8 minutes

approved

wiki

Zoltan: discussed with Daniel to start using the wiki
... how could we put past minutes there

Kaz: there is a way

Daniel: we could also put only the last couple of links

Kaz: tentatively, I can get the links based on the resources on my PC :) and at some point in the future, we could use the automatic tools

Zoltan: I could volunteer to put the scripting repo in order (rationale.md, agendas, etc)

Daniel: the other topic is pre-shared agendas

Zoltan: other TFs and groups do that, so let's do it

Daniel: after the virtual F2F will have 2 weeks vacation, missing 29 June and 6 July.

Zoltan: on 29 June I will be away as well

Issue# 218

https://github.com/w3c/wot-scripting-api/pull/218

Zoltan: considered using uniformly InteractionData with an easy constructor, but I thought it's not worth creating a full stream when it's not used 99% of the time

Daniel: the use of 'any' does not seem consistent

Zoltan: 'any' stands for the value represented by DataSchema
... but we needed InteractionData to be able to represent also streams
... and provided value() as a convenience function for developers so that they don't have to read a stream every time
... most of the use cases are not streams

Daniel: that makes sense
... my only comment was about Actions
... we could make it more clear

Zoltan: I should add a picture to explain in the spec

Cristiano: yes, that would help

Zoltan: will create a separate section, then we can ask feedback from other developers as well

Daniel: maybe just misinterpreted

Zoltan: indeed it needs clarification and we will do it

Daniel: when looking in Web IDL, it is not clear what `any` denotes

Zoltan: when `any` is used, people should look at the algorithms
... but the use of `any` is discouraged for security and obfuscation reasons
... TypeScript is more defined, WebIDL syntax limits us
... we could explain the use of `any` in text, if not in WebIDL

Cristiano: it's pity we cannot express that in WebIDL, but this would be a way out

Zoltan: in web specs, it is the algorithms that define behaviour, we cannot replace it with declarative syntax in WebIDL
... we could also make a practice to use same variable names with corresponding types
... the API is designed for developer convenience, both on ConsumedThing and ExposedThing

Daniel: from the same servient the use for ConsumedThing could be low

Zoltan: it could be used when you federate or proxy other Things

Daniel: question about failures in ExposedThing handlers

Zoltan: the handler algorithms report a JS Error and the impl/bindings use that for generating the protocol specific errors

Cristiano: makes sense

Daniel: will read again the spec and will also reach out to developers, e.g. Christian Glomb

Zoltan: I will make improvements and then I propose to merge it in a draft branch

Daniel: we use the latest published version; the current version is always the draft

Zoltan: ok, will publish it maybe on my github pages, so that it's convenient to see/check it
... will send the link later
... we have the agenda set for the F2F, let's try to merge this PR before that

[adjourned]