scribenick: kaz
<mlagally> If you have a few mins and want to have fun: https://youtu.be/DYu_bGbZiiQ
Lagally: (goes through the
minutes)
... discussion with Chris as well
... any objections to approve the minutes?
(none)
Lagally: approved
Lagally: split into 2 pieces
... call 1: agriculture and transportation
... call 2: lifecicle (including Zoltan and Elena) and health
monitoring
Lagally: (goes through the changes)
Lagally: 8.2 WoT RUntime
... (go back to the changes within pullrequest 475)
... references to docs
... lifecycle diagram
... comparison
... bootstrapping
... would like to merge this
Lagally's diagram on unified device
Lagally: sequence diagrams are useful
here
... bootstrap, onboard, activate, offboard, decommission,
destroy
... 2nd diagram: bootstrap, register, discover, onboard,
activate, use, deactivate, offboard, deregister, decommission,
destroy
... don't want to go into the detail
... but just wanted to show the reource for call 2
Lagally: proposed term definition from the architecture call last week:
[[
Proposal:
A Thing Description Template (TDT) is a description for a class of Things that have the same capabilities.
It describes the properties, actions, events and common metadata that are shared for an entire group of Things. A TDT does not contain enough information to identify or interact with a Thing instance.
(potentially add the following: Note: A TD is an instance of one or more TDTs.)
]]
Zoltan: scripting-specific
Lagally: we have a couple of proposed teminology definitions
Zoltan: subest of TD?
Lagally: yes
... should have discussion during the TD call as well
Zoltan: TD fragments should be also valid TDs
Lagally: right that's not arbitrary
string
... going to write another proposal
... (proposal based on the discussion today, April 2)
... Partial TD = subset of TD that is valid JSON-LD. It
uses
Taki: what would be the whole set
given a "subset"?
... not sure about the concept of "set" here
Zoltan: a document is a set
... should have discussion during the TD call
... we should keep the use case descriptions here, though
Kaz: we might want to look at the
definition of "fragment" from XML-based specs like SSML above
... that implies we might want to have some clear schema
definition for "TD fragment" as well
Zoltan: there is no concrete definition within the spec text, though
Lagally: how about "subse of a TD data model" which doesn't require all mandatory keywords
Kaz: do you mean "all the mandatory keywords of the TD"?
Lagally: right
... (adds clarification for that)
... and "a partil TD is used as a search pattern for
discovery"
Zoltan: maybe we should discuss this
during the TD call
... but OK with this for now
Lagally: ok
... (and save the defitnion as a comment
<mlagally> Zoltan, we cannot hear you
Zoltan: this is just a first
draft
... not sure how to split this up
... a lot of concrete use cases are included here
... public transport by bus, flight cargo, etc.
Lagally: this use case does include
good points
... what would be the expected data model?
Zoltan: this is just a use case
... needs reviewers to improve it
... help from you all would be helpful
Kaz: we can consider this use case as a meta business-oriented use case which inspire us to generate several concrete use cases deribed from this
Lagally: right
... this is a use case on some vertical
Kaz: "vertical" could be a possible term for this purpose :)
Lagally: (adds comments on the discussion so far to the pullrequest)
Zoltan: would like to get a second round for this
Lagally: cargo is ship, airroad, rail
and last mile
... stakeholders are very different companies, however, hared
identifiers for packages, source and target location are
desirable
... shipping options like dangerous goods, priority
Zoltan: would somebody to review this
use case once
... to get better understanding for this
... can mention that within the pullrequest itself
Lagally: ok
... just wanted to capture our points during this call
... assign a reviewer?
... Benjamin for vehicle use case?
Zoltan: sounds good
Lagally: and people on this
call?
... Singapore govtech guys are also expected
... (assigns people from the WoT WG to the pullrequest as
reviewers)
Kaz: note that we're already at the end of the hour
Lagally: ok
... let's talk about the agriculture use case next week
Taki: would like to make contribution to another use case on healthcare
Lagally: great
... aob?
(none)
Lagally: let's continue the discusion during call 2
[call 1 adjourned]
scribenick: zkis
Lagally: past minutes presented in call 1, no objections
[past minutes approved]
<kaz> Mar-26 minutes
Lagally: presenting issues discussed during call 1
Lagally: presenting PRs
... added links to https://github.com/w3c/wot-architecture/pull/475
Elena: have done some fixes to the
diagram
... there was a discussion of it in the Security call as
well
... got input from Oliver, form OPC-UA provisioning angle
... tried to merge OPC-UA lifecycle into the diagram
... Anima focuses on the link between Manufacturing and
Operational state
... Anima also defines the Destroyed state
... maintenance state is not really defined well
Lagally: so it is bootstrapping and
achieving operational state
... any other stakeholder?
Elena: Anima defines bootstrapping
via joint proxy, registrar, and MASA
... Manufactured state has the manufacturer key set
... botstrapping will set Site keys
... also, there are Application keys
Lagally: so it solves the question how to provision the keys during bootstrapping
Elena: it is zero-touch
mechanism
... but you have to have some keys on the device
McCool: not all devices have manufacturer keys
Elena: will get there
... for OPC-UA, the picture is similar
... there is no zero-touch provisioning, the device needs
connected to a network, credentials provisioned
... different requirements for pre-state
Lagally: is there an OPC-UA device without any keys?
Elena: Manufactured state can lack
the keys
... should ask Oliver
Lagally: should we have a security bootstrap state?
Elena: going back to the WoT device
lifecycle picture, we have a separate state for
Bootstrapped/Onboarded
... which is optional
... it is possible some devices are already in bootstrapped
state when leaving manufacturing
McCool: when is the SW installed (behavior)? when application keys are installed?
Elena: will go through the
diagram
... some devices might have manufacturer keys, but no service
provider key or application keys
... then we transition to Bootstrapped state
... configures the device with identity, keys and certs needed
for that
... service provider key is set
... but no application keys at this point
... next, transition to Operational is when app keys are
provisioned and configured
... Maintenance state can update the app keys, but not change
identity etc
McCool: Anima assumes the device is
operational when it has the site keys
... first transition is for service provider bootstrapping,
second transition for app bootstrapping
Lagally: so they have a certain model
for key management
... what is an app key
Elena: from WoT pov, there is a set
of credentials provisioned into the WoT runtime
... this is the application key
Lagally: why ACL?
Elena: the purpose is identification
and authentication
... then ACLs can be set up
McCool: we should separate device provisioning from service provisioning
<kaz> regrets2: Koster
Zoltan: setting up ACLs is needed in OCF for instance, otherwise won't be operational
McCool: we could make a note about device vs service provisioning
Lagally: is there a new state when a
device is registered with a service?
... tried to do some sequence diagrams and had questions like
this
McCool: registering to services is not a state of the device
Zoltan: there are 2 levels of services: protocol native and WoT
Kaz: we need to figure out which
state is linked to which device
... most states can be linked to specific devices
McCool: it might be confusing to
include all possible states
... we can deal with other systems in operational state
... we need to explain what configuration data we are
including
Lagally: that goes to the text, keep it abstract
Kaz: we should think about service layer and device layer separately
Zoltan: operational means the underlying protocol is fully operational, and in addition WoT also works
Lagally: would present own work, then
return to the diagram
... presenting sequence diagrams, 2 flows
... one for the security provider, the other for the
consumer
[link to diagram]
https://github.com/w3c/wot-architecture/blob/master/proposals/unified%20device%20lifecycle.svg
McCool: simple and clean diagrams,
some details to be discussed
... single words should describe states, verbs describe
transitions
... some details like discovery need to be checked
Zoltan: there is discovery for bootstrapping and during operational mode
McCool: words like introduction and exploration could tell them apart
Lagally: this diagram is deliberately not detailed
McCool: we can add details as text
Zoltan: what is the diff between Bootstrapped and Onboarded?
Elena: if we forget about security, define the sequence from operational pov, then we can complete security details
Zoltan: isn't the purpose of the lifecycle diagram to provide input to Security?
Elena: it has its meaning even without security
Lagally: I agree. Even if we have a secure environment, this diagram should work
McCool: I like that each state
represents a new relationship
... we could start there and look it from the actors point of
view
... we should define things from the perspective of
relationships
Lagally: we could make a table with state and description and relationship
McCool: depends on what relationships we aim for
Kaz: identify the actors
explicitly and a group of states would be useful.
... we could use the colors to idetify them. also possibly assign soe of the
states as a meta state linked to oprational state.
... anyway, thinking
about the relationship (or mapping) between Michael Lagally's simpler
diagram and Elena's detailed diagram would make sense
Lagally: about next steps: will create a table, will update the diagram, Elena can take it from there
Elena: would not like to take it
from an empty table; for security need to have the states and
relationships, capabilities etc defined
... everything that is not security related
Lagally: OK, let's work on the table first
McCool: in github, preferably, e.g. in an issue
Lagally: created https://github.com/w3c/wot-architecture/issues/476
any other topic?
<mlagally> Just a little bit of fun: https://youtu.be/DYu_bGbZiiQ
AOB?
[call 2 adjourned]