<scribe> scribenick: zkis
<kaz> https://www.w3.org/2020/04/20-wot-minutes.html
Zoltan: any issues with that?
no objections, minutes approved
Zoltan: DP mentioned talking with people about API design
Daniel: yes, people preferred a
symmetric API, but it's a compromise
... there was going back and forth
... regarding the issue discussion, ZK's latest proposal is
good since it extends the API unbroken with new
functionality
... so the "any" sometimes could be InteractionData
... agreed with Ege this is a good start to explore this idea
in code, without changing the API
... so it's good solution for the moment and in the simple case
there is no overhead
... one can provide a simple integer for instance
Zoltan: yes, that was the idea
... and it also forced me to define the algorithms for
bindings, which we have not done yet
... but it turned out to be good
... we can change later to InteractioData only, or keep it this
way
Daniel: maybe we could merge the PR but create a new issue that captures the discussion
<kaz> Issue 201
<kaz> PR 209
<kaz> section 6.16 ConsumedThing Examples (within the PR 209)
Zoltan: there is another possible solution; we always assume simple data and we define a "content exception" that would be handled via InteractionData.
Daniel: wondering how would that work
in code but the separation might be useful
... we could also plainly fail and then in another call recover
with InteractionData
Zoltan: that's a good idea
Daniel: but not sure that's a scenario what people would use
Zoltan: what about passing an app option to read to tell implementation how to present the data?
Daniel: that makes sense
Zoltan: another thing is, should we
have options for language and encoding for reads?
... that would allow presenting a string that has encoding and
language in a simple way (now it is only possible via
InteractionData)
Daniel: there was a language
negotiation thing in the TD
... currently the implementation would take the locale settings
as default
Zoltan: it would be a scripting
specific SW option, making app's life easier
... also, doing it in implementations is safer
... we have another option: always use InteractionData but put
there the simple value as a property
<kaz> 7. The InteractionData interface (within PR 209)
Zoltan: will make a comment to illustrate the idea
Daniel: will discuss that with Ege and Sebastian
Zoltan: we need to experiment by writing code examples and see what works best
Daniel: node.js now supports
WASM/WASI
... from v14.0.0
<dape> https://nodejs.org/api/wasi.html
Zoltan: WASI doesn't yet support network operations
MM: the use case is to use something compiled, to run on a Scripting runtime
Zoltan: TypeScript is supported in
WASM
... the problem is not every system call is implemented in
WASI
<McCool> https://wicg.github.io/web-transport/
adjourned