<scribe> scribe: zkis
<kaz> Sep-14
Zoltan: any objections for accepting the past minutes?
none
Past minutes accepted.
Cristiano: some questions, the problem
is that we are defining the Scripting API in 2 different
ways
... one is defining a global object
... other times we accept a namespace that is not global
... the problem is this doesn't work as it's not recognized as
a namespace
... we need to change the TypeScript definitions
Zoltan: and also the Scripting API?
<kaz> index.d.ts
Cristiano: we should find a way to make the TS definition work in the browser as well
Daniel: we had a discussion on node-wot and all proposals had some problem
Zoltan: do we have a Scripting issue for this?
https://github.com/w3c/wot-scripting-api/issues/215
Zoltan: relevant opinion:
https://github.com/w3c/wot-scripting-api/issues/3#issuecomment-283746764
... so, 1) the root object should not be constructable
... 2) if it has no state, then a namespace is OK
... 3) if it has state, has to be a global object
... we also have conformance classes, we can group along
those
Daniel: there was a lot of discussion since the workarounds had issues
Cristiano: the recommended way is not to
define a global variable, but use a namespace
... since making a global variable will duplicate the
functions
Zoltan: we should also keep in mind future functionality like management interface, that could be a new conformance class, namespace or global object
Cristiano: if we export as namespace
WoT, it worked
... we don't have a global variable in node-wot
... we could have 2 TS definition files
... or if we have one TS definition, we need to explain this vs
modules/namespace/global object
Daniel: we'd like that everyone could use the same TS definition
Cristiano: the problem is that it works if we create a runtime, but doesn't work from inside a runtime
Zoltan: subnamespaces?
Cristiano: not supported in TS
Daniel: if we just declare the namespace WoT, would it work?
Cristiano: we don't have a module
then
... so we cannot import it as it was a module
Daniel: could the implementation manage the module vs implementation transparently so that code like WoT.consume() works?
Cristiano: for short term, I could make a PR to test with node-wot and trying to make it work
Daniel: we could have a playground
for testing these ideas on npm
... we could have a snapshot to play with, in a branch
Zoltan: if there are effects on the Scripting API, please update the issue
Zoltan: we need to clear the open issues
Daniel: we probably cannot close all issues
Kaz: publication for FPWD/FPWGN does not have to be perfect
Zoltan: FPWD or a Note?
Kaz: First Public WG Note if we
change the name and the shortname URL
... we could also change the name to e.g. Scripting API 1.1
Zoltan: I would vote for just updating the current publication
Daniel: agree, no jumps just evolutionary change
Kaz: there will be several
difficult points all the time
... the suggestion is to go ahead and publish
Daniel: yes, and there will be
remaining issues
... IMHO all Web IDL related issues should go in
... for instance the current PR (#264), but improving the text
could be done later
... so I would not postpone it long, because it needs to be
implemented and tested
Zoltan: how long we need for publication
Kaz: a week of group review and addressing feedback
Zoltan: it means we need to complete all changes by the group call on Wednesday
Kaz: got it, but note that there will be no group calls during F2F. also probably it would be difficult for people to review the updated draft during the PlugFest next week too
Zoltan: there is a pretty new API to check for idioms, https://wicg.github.io/cookie-store/
Daniel: have a question for multiple consecutive optional args
Zoltan: should work
Cristiano: I agree, it should work
Zoltan: we need to check it for
TypeScript
... dictionary and function are nullable so it should be
OK
... we also want to create an issue for defining PropertyMap as
record instead
Daniel: the Web IDL change looks good
Cristiano: looks okay, are we ok to create the new pattern with Subscription
Zoltan: yes, it is ok. Also check the Cookie Store API for when to use an event and when a subscription function
Cristiano: question about unsubscribe Form
Zoltan: raised that issue in the TD
TF and asked for help in the group call
... now we have a proposed algorithm, but could be improved
Cristiano: until when can this PR be commented on?
Zoltan: I would make a constraint for the TD to defined unsubscribe Forms together with subscribe Forms
Cristiano: would be good design but
would break the TD spec
... we need to discuss this in the TD
... we should issue a separate issue for that
Zoltan: right
Daniel: like a generic "how to find an unsubscribe Form for a subscribe Form"
<scribe> ACTION: ZK to file an issue at the TD spec
adjourned