Meeting minutes
Logistics
The next meeting will be April 28
McCool: I won't be as engaged with WoT as my focus for W3C is changing.
I can probably make the call twice a month
We may be able to interest people in the Smart City group, Kaz can ask...
Sebastian should be able to help out
Minutes
<kaz> Mar-10
McCool refers to his use case
Some discussion on north bound APIs in relation to TDs
We approve the minutes.
Activities
(for 14 April 2025)
<jhierro1> Also Martin prepared a use case in form of PR. He will join later today
Issue 18
An outstanding question is how to map actions
McCool: I'd like to walk us through what we mean by actions
<kaz> Issue 18 - Discuss Action Semantics
We have an opportunity to align on what we mean by actions
Juanjo: actions are not currently supported in NGSI-LD, but they are on the roadmap
<kaz> WoT Thing Description 1.1 REC - 5.3.1.4 ActionAffordance
McCool: let's look at the current TD spec. This describes how to invoke a function, which may or may not change the state
<jhierro1> but this is not a problem but an opportunity because actions then can be defined in NGSI-LD in a way keeping an eye how things can work together with WoT.
A non-state changing function might for example change the units of measure
The vocab term "safe" refers to whether an action leaves the state unchanged
idempotent actions are guaranteed to have the same effect
… each time they are invoked
Synchronous actions signify that any state change is complete by the time the action's response is generated
That isn't the case for async actions
Actions further define the data schema for their input and output
[Martin joins, and Michael summaries the discussion so far]
Actions can be invoke functions or they can initiate processes
Juanjo: the way to invoke an action should be the same for all actions whether or not they change the state etc., right?
McCool: yes indeed
The affordance in the TD tells you how to invoke the action
McCool: synchronous actions may take appreciable time just so long as action is completed
<McCool> w3c/
Kaz: who is durkinza? Please type your name into IRC, thanks
Zane: I am in the Web of Things activity
PR 17
McCool: my use case was about onboarding
Martin's use case is how to describe an NGSI-LD entity as a thing
<kaz> PR 17 - Bauer Thing Description for NGSI-LD Entity as Thing Use Case
He can use NGSI-LD to model digital twins
McCool: WoT TD's can't describe all possible ways to use HTTP
We can use feature requests to discuss any specific examples
<kaz> rendered MD
Martin: events should be easier for us to map
Another example is where you have commanded a robot arm to move, and want to track its progress
Martin: we need to look at some specific examples to see what problems arise
At ETSI, we are still considering the use cases ...
Kaz: we can look at some concrete examples to elucidate the requirements
PR 14
<kaz> PR 14 - Define Onboarding Use Case example
McCool: we need to talk about stakeholders at some point
I want to merge the outstanding PRs, any objections [no]
Stakeholders
McCool: The WoT threat model categorises stakeholders
… but not Digital Twin modellers
Martin: We don't yet do that in our work
McCool: whilst we hav a list of stakeholders in the WoT use cases, we don't yet provide definitions of what the terms signify
<McCool> w3c/
Examples of mapping between NGSI-LD Entity and WoT Thing
McCool: we can try to write a TD for a given NGSI-LD entity, I will create an issue to track that
McCool asks for some candidate NGSI-LD entities for this exercise
Martin: 2 or 3 examples, e.g. a concrete device and something more abstract
McCool: we would then identify the gaps, i.e. things that can't be modelled in TDs
… and capture feature requests for NGSI-LD as issues
Our next call is in 2 weeks from now, and it would be good to have some examples in progress
<kaz> Issue 20 - Implement Test Case for "Entity as TD" Use case
Some discussion on the existing use cases ...
Juanjo: we need a use case involving collections of things
Next meeting in 2 weeks
<kaz> [adjourned]