<scribe> scribe: zkis
past minutes approved
Daniel: MMc also asked for approving the F2F minutes
<kaz> monday minutes
no objections, minutes approved
Daniel: on the TD call the issue was acknowledged and more work is needed on the Event affordance
Ege: we need some ideas from existing experience, e.g. M. Lagally
Zoltan: does it make sense to expose subscribe id's or keep them private in the impl?
Ege: usually not, except perhaps
in few use cases
... subscription payload might make sense in certain
application
... maybe we can say if there is a use case, we come back to
this
Zoltan: so until then we identify events by names in Scripting
Daniel: the implementations needs to
keep them as a state
... today we don't expose the subscriber id's
... we don't fully support yet any filtering on events
Zoltan: do implementations understand
subscription data or is it opaque?
... in the current bindings?
Daniel: at most we pass subscription data along in node-wot
Zoltan: the Oracle use case would need that node-wot passes on the data
Daniel: yes
DP made a semantic layer prototype in https://github.com/danielpeintner/thingweb.node-wot/commit/844e199357d84494030549dadd655d20fe8a0706
Daniel: this is a layer built on the
top of the existing Scripting API
... enables fetching e.g. a semantic Property
... supported for ConsumedThing
... this is for experimenting
Zoltan: I like it's a layered approach, like a library; not sure we should standardize yet
Daniel: it would be good to standardize
Ege: would not oppose this, and should be standardized
Zoltan: the standardization could happen in a new spec, like Generic Sensor vs Accelerometer specifications
<kaz> 6. The ConsumedThing interface
Zoltan: it could be separate conformance class
Daniel: it makes sense to do it in
other spec that extends the current API
... we should get M. Koster and other and experiment with it in
plugfests etc
... for instance about parameters etc
... we should look into semantic extensions
Zoltan: what are next steps? plugfest PoC?
Daniel: will work on implementation in node-wot, but need input and feedback from M. Koster and others
Zoltan: so first part of the work will be in node-wot
Daniel: there will be the plugfest call, any plans for the virtual plugfest?
Ege: there was the VPN tool
proposed and needs to be tested
... so that we could make virtual plugfests easiers
<kaz> Mar-25 PlugFest minutes
Zoltan: should we introduce a
Scripting-specific data description type that would merge
mediaType (contentType from Forms) and DataSchema?
... otherwise in order to extend the algorithms, we'd need to
define everything the TD spec defines
Daniel: going with Value would be
fine, we don't even need DataSchema there
... it's repetitive
Zoltan: that is true also for mediaType
Daniel: we need the mediaType because
we don't know which form was used
... app can look just at the mediaType
Zoltan: we can get DataSchema by property name from the TD
Daniel: right
... we need this both on input and output
... we still need the formIndex in InteractionOptions
Zoltan: I can create an experimental PR for this
Ege: makes sense to me, too
Daniel: the only fear is to break existing implementations
Zoltan: that requires some
thinking
... should we include API version in the spec?
Daniel: never seen a good solution with a version
Zoltan: let's continue in the issue discussion
<kaz> [adjourned]