<scribe> scribe: zkis
<kaz> Feb-17 minutes
Past minutes accepted
<kaz> PR 203
Zoltan presented the PR changes: added formIndex as a hint for implementation, and clarified the use of write handlers
Zoltan: the PR has been reviewed,
approved and merged
... it fixed two issues, 199 and 202
Daniel introduces the issue #200
Daniel: would say the table in Ege's comment should be informative
Zoltan: agree that it should be
informative
... we can include this into the algorithms
McCool: we need to define what is the behavior if a non-mapped error happens
Daniel: some errors indicate if an
unspecified error occured
... showing "UnknownError" as an example
... in the basic use cases we do need a clear mapping
... but there might be error that could be classified in many
ways
Zoltan: what about the error data?
Daniel: we could use the error code,
but no lengthy error messages or error data
... up to the implementation to harden that part
Zoltan: nevertheless, the spec should have some guidance on what data to include with errors
Daniel: is there an issue with privacy if we don't specify in the spec?
McCool: we should maybe have guidelines and good practices, but implementations can decide to follow them
Daniel: need to check node-wot in this respect (what data is exposed in errors)
Zoltan: we can make a note until we know exactly how to curate the error data
Daniel: let this issue be the single
place to contribute to the error mapping
... or create a PR?
Zoltan: I can create a PR that can be left open, in order to experiment, but the primary place to discuss and contribute would be in the issue
https://github.com/w3c/wot-scripting-api/issues/201
Daniel: we have an open PR in the TD spec that is related
<dape> https://github.com/w3c/wot-thing-description/pull/869
Zoltan: we could still describe the
current way of how things are supposed to work
... and use the information from the PR
Daniel: that is right; we should link the PR and the issue
https://github.com/w3c/wot-scripting-api/issues/193
Daniel presents the issue
Zoltan: there was a question which Form would be used for reporting the written value
McCool: here as well it applies that
implementations define the behavior and the TD should describe
interactions correctly
... a safe assumption is to say writes never return a value,
and you need to define an action if you want a value back
Daniel: seems there is no consensus on this
Zoltan: you can't figure out precision from the value (assumingly) reported by the write
McCool: that would need an action with a defined output schema
Zoltan: I agree
McCool: I agree with the current spec that returns Promise<void> because it handles all cases correctly
Daniel: looks like the discussion goes into the direction to keep the current way, because it gets tricky how to do the alternative correctly
McCool: we need to log a design decision and an editor's note, in order to avoid duplicating the discussion
Daniel: one possibility is to record the decision
McCool: we need a separate document to summarize the key design decisions
<scribe> ACTION: Zoltan make a PR with a note and explanation, and update the explainer
https://github.com/w3c/wot-scripting-api/issues/204
McCool: I propose to discuss this in
the Discovery TF
... and we would track Scripting related aspects in this
issue
AOB?
none
[adjourned]