Meeting minutes
Agenda Review
<kaz> Agenda for today
Ege: registry, queryaction issue, opc ua uri scheme
Minutes Review
Ege: any objections?
Ege: minutes are approved
Binding Templates
OPC UA URI Scheme
eclipse-thingweb/
Ege: it is specific to OPC UA but can be generalized
… opc ua spec does not prohibit special characters
… we get `=`, `;`, `"`, and so on
Ege: we can encode everything but it is not readable and not copy pastable
Ege: sebastian brought the opinions from the OPC UA wg with two options
… one is to define query params and the other is to use another term
… in both cases, " needs to be escaped
… the first one is annoying for copy pasting parameters from OPC UA tools and the second one is not compliant to our href design decisions
Cristiano: I lean towards option 1 since it aligns with WoT more. It doesn't help opcua developers
… it would be a browser tab open that you copy and encode
… I value consistency
… it is like the MQTT discussion
… is the `"` necessary?
Ege: yes you need to send it over the wire
Daniel: namespace can need url encoding too since it can contain `#` too
… why are there multiple query parameters and why not everything is encoded?
Daniel: you cannot know that it is encoded?
Cristiano: you can always put it into the decoder and get a decoded url
Luca: repeating cristiano, the element that is there is a URL. Everything that is specified for the url applies
… hopefully initial connection can remove some of the redundancy
Kaz: the core is about URI in the end. We should ask TAG about this as well
Ege: how can we start the discussion with TAG?
Kaz: also we should mention W3C WoT TF in the comment, because this is not our own issue but an issue on the ecripse-thingweb's node-wot repo
Kaz: writing an issue that summarizes the problem would be good
Ege: ok will write it after the call
Binding Registry
Ege: please express your opinions on inclusiveness at the issue w3c/
Thing Description
Issue 2068
Ege: this came up in the scripting api but it actually applicable to more
wot-thing-description Issue 2068 - Specify the result of queryaction operation
Ege: cristiano proposes defining another term
https://
Ege: in my proposal, I have also proposed new terms with input and output inside them
… however, all these proposals go towards TD 2.0 but we need a solution for TD 1.1 too, probably
Luca: it would make more sense to deal it in the scope of a topic like "relationships between affordances" or "manageable affordances"
… this is indeed a sore point
… so the main question would be is whether we want to add new terms like this or should we have relationships between affordances, like reading a property giving the query result
Ege: It is a bigger discussion indeed. We could have many such terms coming if we go this direction
Luca: we should review the what affordances really mean, like is an action a queue all the time
… what is the boundary between an action and a property
Cristiano: TDs can be used to describe AI agents. We need to figure out what to do right now
Cristiano: this is like a bug in the td 1.1. We seem to have forgotten to add a field like 'status' here
… we can say we don't know the payload or use the profile
Ege: additionalResponses can be used here but it is a bit of a hack
Zoltan: I am active in agentic workflows, like composable microservices. We can start with something simple
… the result of query is an object. So we can do multiple operations like invoking, canceling, querying
… I would not want to go for recursive mechanisms though
Ege: we can look at https://
Zoltan: we should define what use cases we are resolving with action affordances
… we should start simple
Kaz: Given the time left, we should clarify the pain points and the use cases
… and then we can find a solution
Zoltan: we should clarify what level of status is delivered to upper layers in the application
… so the semantic status and the protocol level status
Kaz: we need more time so let's discuss it next week too
Cristiano: I agree that we can discuss next time
Ege: make sure to comment in the issue
Ege: AOB?
<kaz> (none)
Ege: adjourned