Meeting minutes
Guests
Ege: as a guest today we have Luca Barbato
… he is in the process to became an Invited Expert
Agenda
<kaz> Agenda for today
Ege: we will discuss two main topics ids in TD and Registry for protocol bindings
Minutes
<kaz> Feb-12
Ege: do we have any remarks?
… anything to change?
… minutes approved
ID optionality
<kaz> Issue 2054 - Reconsider id being optional
Ege: the whole point is to make id mandatory again
… there has been multiple opinions
… coming from implementers
… in the case of node-wot a random one is generated
… other people used similar strategies (filtering out TDs without IDs )
… in the past this topic was controversial as we had to do two CR
… implementers gave positive feedbacks for having a mandatory ID
… we can make it mandatory but "not static"
… main reason why we make it optional was that the ID could be use to track people associated with devices.
Kaz: good summarization, we should think about different levels of ids for IoT systems
… potentially we might use encription in some use cases.
Cristiano: thanks for the summary. I am ok to make it mandatory. Not making it permanent needs guidance for the implementers
Cristiano: should it change every hour or so?
Ege: I think it should change per network or installation
Luca: I agree with the summary
… we should start to see if have consensus to build our own namespace
… right now if we want to use that we have to pick one of the specific namespaces that are avaible
… we don't have anything that is actually really fitting
… it would be nice to have a short namespace right after urn:, example urn:wot
Daniel: we should reach out to the security and privacy group avoiding to change back
Ege: the idea is to have WG consensus first
… then ask them for feedback
… if we get a greenlight we proceed with a PR
<luca_barbato> https://
Kaz: recently we got new Team Contacts for the Privacy WG and the Security IG, so we can start to talk with them too. In addition, technically, there is a possibility to use DID as a potential ID for Thing Descriptions
Ege: should we register a schema in IANA?
Luca: yes
Ege: for DID we should register a DID method
… for a IANA we have first to create a section in the specification and then we should be ready to register it.
Cristiano: I like the idea of registering DID method or URN namespace
… what is the advantage of registering the namespace at IANA or DID method?
… so creating something ourselves
Luca: now id is any kind of URI
… having a constant namespace would make the implementers life easier
… and also easy to introduce for us
… we have options to format our own urn schema
… it is basically suggest to TD designers to use a fixed format
Kaz: Technically, we as WoT WG can define this kind of notation and registering it to IANA but I'm a bit skeptical because it would result as something like DID.
Ege: I see what you mean
<kaz> DID Core spec
Daniel: I second what has been said, we can provide guidance about one can or should use. however I'm strongly against to mandate one single format.
… I would not force any particular format, but we can register our own.
Cristiano: if we force the use of urn or did, we lose the use case of using the url where the TD is fetched in the first place
… so enforcing a format will block doing that
… maybe he is already using the id as a URL so it can break implementation
Luca: we are not proposing to change what is the ID, we are just saying that is mandatory.
… I'm still okey to leave the id string to be any kind of string
… I'm still okey to leave the id string to be any kind of URI
… I would just add some additional guidance (which is needed from the implementation pov)
Kaz: Luca is correct, the original concern of the privacy group was to use Global unique ids
… so we should answer what kind of IDs to be use and how to use it
Ege: ok so we have consensus on 3 points: id is mandatory, is not globally unique or permanent, it should not be in a specific format but guidance of which format to use is needed.
Cristiano: about point 2 this means that I can't use UUIDs?
Ege: well if they are not permanent then you can
Kaz: is it a should or a must? I personally think we should say "SHOULD" here at the moment.
<luca_barbato> +1 for SHOULD
Ege: in my opinion it should be a must
<dape> +1 for SHOULD
Ege: if it is a SHOULD we can still encounter privacy issues
Luca: but there might cases where you cannot get away with immutability of the ID
Binding Registry
Ege: we have split all the controversial topics in sub issues
… the one that are ready to landed have the PR needed label
… we have two PRs ready
PR 411
<EgeKorkan> PR 411 - Handling Conflicting Submissions
Kaz: All the sub issues derived from the questions within the registry requirements doc have the "registry mechanism" label. Right?
Ege: yes.
… any other comments?
… ok PR approved and merged. The document is still in the draft status
PR 412
<kaz> PR 412 - Adding Summary Document Requirements
Ege: this is about summary document
Ege: any remarks for this PR?
Cristiano: does the reviewer have access to the real binding document not only this summary?
Ege: yes
Ege: PR accepted and merged
[adjourned]