<inserted> scribenick: zkis
<scribe> Agenda: release discussion, updates, discovery, implementation, testing, next calls.
Daniel: any other agenda points?
<kaz> July-20
Minutes approved
Daniel: People should take a look at the spec but people are on holiday. I thought about end of August.
Zoltan: I agree
Daniel: we should give 3 weeks for checking the spec
Zoltan: we can fix the ReSpec error
by checking the cross-references
... end of August sounds good
Cristiano: for me too
Mizushima: I am on holiday in the end of
August
... but no problem
Daniel: will write an email asking feedback until 24 August
Daniel: we should defer this removal
a little more
... we can put a note these features are in danger
Zoltan: we can also decide to include only stable features in the spec, so remove it and re-add it when it becomes clear
Daniel: node-wot already has it and
if people use it then we should give some time
... the TD task force has the same issue
Zoltan: we need a deprecation mechanism in node-wot
Cristiano: we could mark it in the
TypeScript definition
... what about of differing from the TD on this?
... but we should synch with the TD spec
... the API version should be linked to the TD version
Zoltan: we could include properties with the API version and the TD version
Cristiano: we could also include it in
the API doc
... so not only machine-readable, but also human-readable
Daniel: Cristiano please create an issue about this
Zoltan: does the TD spec have an official version number?
Cristiano: the spec has one
Zoltan: the TDs have a version, what is the rule there?
Cristiano: is that the ontology version there?
Daniel: we need to check that
Zoltan: vocabulary and ontology versions should define it
Cristiano: right, we need to discuss this in the TD call
Daniel: maybe there are other changes as well
Cristiano: as a developer I should be able to point to a page or a version
Zoltan: optimally we should be able to define a dependency in Node
Cristiano: will create an issue at the TD and also at Scripting
<scribe> ACTION: Cristiano creates issue for versioning on the TD and Scripting specs
<kaz> July-20 Discovery minutes
Daniel: who was present on the call?
Cristiano: I was, taking the
minutes
... maybe we should check the minutes
... presents the discussion from the Discovery call
... in the first version of the spec there are sections which
use some conflicting words vs our usage of the same words
... we discussed that with Zoltan in github issue
... they use 2-phase discovery, 1. find the service, 2. use
that service
Zoltan: the current API reflects IoT protocols discovery mechanisms, but not the WoT-specific 2-phase discovery
Daniel: the first phase is what is not supported in the current API
Cristiano: authentication is also in the loop
Zoltan: which should be pushed outside the API if possible at all
Daniel: we do have an issue to conveniently fetch a TD, which corresponds to the "direct" discovery
Cristiano: in the Discovery TF usage of the word direct refers to Directory
<kaz> WoT Discovery initial draft
Cristiano: we need more work on this, so we cannot make the Scripting changes right now
Zoltan: I agree, especially with phase 1
Daniel: so Cristiano and Zoltan are checking the Discovery TF works
Daniel: if we publish the spec, we
also need to update the implementation
... hopefully could find the time in August for this
... we need some help with testing
Cristiano: is that full end to end testing like Ege is doing, or just unit tests?
Daniel: I very much like the test
suite that has a server running and will test a client against
the API
... and gives a mark for the coverage
Cristiano: that scoring mechanism would be nice indeed, but it's quite difficult IMHO
Daniel: the issue number is 190
https://github.com/w3c/wot-scripting-api/issues/190
Daniel: it is difficult because we need both a server and a client
Cristiano: how do we test that a client
consumes a TD in the right way?
... how do we check the correct behaviour of an
implementation
Daniel: it's doable in JS and TS, but in a different language it's different
Zoltan: we could define a test spec
and implementations could implement that
... in language-specific way
Cristiano: that would be nice, needs a lot of work
Zoltan: the algorithms in the spec should be clear enough that a test suite could be created from that
Cristiano: was thinking about the same
point
... we could also refer to a reference implementation such as
node-wot
Zoltan: so when in doubt how to implement, please check node-wot?
Cristiano: yes, something like that: check the spec and the reference implementation
Zoltan: indeed, and implementations
should give feedback on the spec and also on node-wot
... I think we should first create tests in node-wot, and then
see what can be generalized
Cristiano: should be like an integration
test
... I wrote some tests like that on the Java
implementation
... of another spec
Daniel: I will check with Ege
Cristiano: while we are implementing the new spec version, we could also think about how to test
Daniel: right
Daniel: on August 3 neither Cristiano nor Zoltan
is available
... so we skip next week
Zoltan: after August 10 it's fine for me
Daniel: I could give availability later
<scribe> ACTION: Daniel to send email for asking for spec feedback
Daniel: any other business?
adjourned