W3C

WoT Scripting

01 March 2021

Attendees

Present
Cristiano_Aguzzi, Daniel_Peintner, Kaz_Ashimura, Tomoaki_Mizushima, Zoltan_Kis
Regrets
-
Chair
Daniel
Scribe
cris, dape, zkis

Meeting minutes

previous minutes

<dape> Feb-22

Daniel: we looked at the f2f wiki, we have to improve it. Maybe discuss again the time frame
… we merged the a pr from Cris even if it could be improved with more examples
… then we reviewed issue 302. We decided to not update WebIDL but only TS.

Zoltan: btw InteractionInput is any so that's why WebIDL update is unnecessary.

Daniel: yeah but I prefer to have WebIDL and TS aligned.
… then we reviewed 293 issue. We have two open PR to solve the issue. We'll discuss later.
… minutes ok?

Zoltan: yes

Daniel: ok minutes approved.

Quick updates

<dape> F2F wiki

Daniel: about F2F. I modified the topics on the link above
… the group should provide feedback on that points.
… then there's the rename to master branch to main. Other groups are converging. Not big issues just some warnings when doing the update.
… anyhow let's postpone it

PR 305

Daniel: Factor out maplike interfaces property reads and writes - https://github.com/w3c/wot-scripting-api/pull/305

Daniel: Zoltan changed PropertyMap to a maplike object.

Zoltan: it has utility methods like forEach

Daniel: it's fine by me.
… cristiano also approved it

+1

Daniel: next pr:

PR 308

<dape> PR 308 - Introduce PropertyReadMap and PropertyWriteMap in TS

Daniel: is the counterpart of the previous PR, but only for TS

Daniel: it's ok but there are some code style issues.

Cristiano: I'll fix the issues after the call.

Daniel: so let's merge after the call
… any objections?
… ok then I'll also publish and increase the version number

PR 307

PR 307 - fix 302

Daniel: it's about the return type of ActionHandler and invokeAction
… it propose to use undefine.

Zoltan: are you sure that in TS needs to declare that something can be also undefined?

Daniel: not sure

Cristiano: I remember some errors in other expirements that I did. I'll have to check.

Zoltan: ts is more strict

https://www.typescriptlang.org/play?#code/GYVwdgxgLglg9mABFApgZygCgJQC4DyARgFYrSIDeAsAFCL2IBOKUIjS4AJisDGCpwDctAL5A

Cristiano: it seems from the link above that you should declare it as undefined or object

Daniel: so we don't go out of sync?

Zoltan: no, as long as the algorithm is clear.

Daniel: so do we agree to merge?

Zoltan: ok

Daniel: merged

Issue 306

Issue 306 - Feedback about InteractionOuput

Daniel: let's start with 306

Daniel: it was originally started by sebastian.
… we were discussing about having the new api or old apis in this plugfest
… he wondered how the new api will look like
… he was not sure that the new api is intuitive
… to summurize, the old api for the simple case was good, but new api can cover more usecases

Zoltan: the code in the issue it is not really correct. It assumes that you know that a property will return a json value
… for a complete picture you should add every other bit of code for error handling
… because the example hides other details

Daniel: that's true even with the new api

Zoltan: true
… btw in the new api you know immediately where you're failing.

Zoltan: there are a couple of option on the issue body
… thenable it seems the more promising
… but it might not super uderstood
… and the only benifit is to remove one await.

Cristiano: thenable are fine, but not sure too if it is really worth
… value as function as also benefits from TS world. You can define type guards

Daniel: I think this is a precious feedback from developers, we should pay attention to it.

Zoltan: stream can handle all usecases
… we can look at some examples

Daniel: currently the runtime decide to provide a stream or a value

Zoltan: no it's the application to decide

Cristiano: true

Zoltan: if dataUsed is true the application should not use stream interface

Daniel: if you call value arrayBuffer will fail, should we improve it

Cristiano: yes it might be worth doing it.

Zoltan: we don't really have an use case for this
… it is up to us anyway

Daniel: it blows up the memory true

Zoltan: if it is confusing we could remove it
… it is equivalent to a readable stream
… you are rising an other usecase: buffer is similar to value.

Daniel: I think to have the benifit to having arraybuffer or Readable stream
… do we know any devices that does not have Readable Stream?

Zoltan: zephir os is an example

Daniel: the consensus seems to be that the current API is in a good shape.

Cristiano: I am good to close it

Daniel: closed

Daniel: time is up

<kaz> [adjourned]

Minutes manually created (not a transcript), formatted by scribe.perl version 127 (Wed Dec 30 17:39:58 2020 UTC).