See also: IRC log
Minutes have been approved
<renato> https://www.w3.org/2016/poe/track/actions/48
victor: Has only just read the email. Has been updating the list of normalisation steps.
michaelS: The Json-LD issue: There is a
strange display of blank notes. I tranformed ttl into Json-LD.
... The result had "undefined" in some places
<victor> +q
michaelS: but it worked on the playground so
I raised an issue in GitHub
... turns out there was a known issue in latest release (it delivers
blank nodes not nested as we have it) - with advice on how to fix
<victor> The nesting problem can be seen with the examples at http://odrlapi.appspot.com/. This page shows the results of the automatic transformation
benws: is the nesting issue display or critical?
michaelS: icing on the cake
ivan: lets close the issue
... we cannot always rely on other tools to create the display we want
benws: issue 49 - I will write commentary
<renato> hang on....
<victor> ok
<michaelS> https://www.w3.org/2016/poe/track/actions/49
<renato> https://w3c.github.io/poe/model/#metadata
<simonstey> must retrieve and process is always a bit problematic
renato: should we get the policy pointed to, or just stop processing
victor: says we don't have an action we can do
<simonstey> we had the same issue with profiles
renato: when you first get the policy its not in your knowledge base - do you add it or do something else?
victor: se we should get it and add it?
renato: perhaps get the new one and replace the current one with it
<simonstey> then we wpuld have to describe how "getting" works
benws: so does it invalidate the permissions in the old policy
<simonstey> or what should happen if getting fails
<renato> https://www.w3.org/2016/poe/wiki/Validation
renato: this is about parts - we've added it - asks victor to confirm
victor: changes need to be made as per
emails sent
... for the party collection
michaelS: so new test cases should be added to report?
yes
benws will add documentation. Some concepts seem to be in that shouldn't be - but I am trying to understand further
scribe: question about multiple assets, multiple parties
<michaelS> https://docs.google.com/spreadsheets/d/1I2-qht3KRjkIvwdvsfkAMq4AJ0FXjn3Wl5mH-tPe5bs/edit?usp=sharing
victor: it would be possible but validation would be difficult
michaelS: this is for implementors. walks us through it
<michaelS> https://iptc.org/temp/ODRL-CR-ImplementationReports.html
michaelS: and I have created some python for
the HTML output
... should put it up somewhere
renato: so we can make the google doc public soon
<renato> https://w3c.github.io/poe/test/implementors
<michaelS> Short URL for the implementation report - please use that one: https://goo.gl/iXnonW
renato: implementors on call should add their names to the doc
ivan: must remember to fill the missing tests cases before publication
<renato> https://github.com/w3c/poe/issues/267
renato: summary - the consequence property
has to be satisfied if duty is not satisfied, but later we say its an
*additional* duty and the original duty has to be satisfied
... michael pointed out that the original duty may sometimes be
unsatisfiable (e.g. missing a date gone past)
... need to get group agreement on actual intent
sabrina: I see it as - there are duties and
there are consequences if they are not fulfilled
... e.g. you are fined if you don't get permission of a data subject
... paying the fine doesn't mean things are ok necessarily
... I think consequences are necessary but its not clear how they ;ink
to the duty
benws: so fulfilling a consequence doesn't mean the duty is complied with
sabrina: even if consequence is complied with, the original duty still holds
renato: that's what I meant to say
<simonstey> its kinda the same as with remedies
<simonstey> you are still prohibited to do xyz
<simonstey> even if you have fulfilled the remedy
sabrina: agrees
renato: so duty must be satisfied as well as constraint
michaelS: approach is correct
... but how do we write this down using odrl?
<renato> https://github.com/w3c/poe/issues/275
michaelS: returns to issue when time limit
is specified for a duty but has gone past
... do we say - evaluation of each duty is needed
... many duty actions may be expressed by users
ivan: is this an editorial change or more?
michaelS: I think its that we have contradictory statements and we must choose one to follow
benws: is the issue resolved by the clarifications from serena and renato?
michaelS: yes, but..
... current processing rule doesn't work well for me
renato: so first part is editorial - we can just add "also"
renato: other issue is how we say that a
duty *must* be satisfied even if it can no longer be
... for the IM to come up with all possible ways to satisfy duties in
the future will be tough given all the different scenarios that people
will come up with
benws: agrees
... we can't build a logic to describe the world
... lets keep the simplicity of odrl as now
... and handle this as an implementation issue
sabrina: agrees. this has happened in other
cases.
... generally people change the duty to make it satisifiable
... the application should highlight if something is not satisfiable
michaelS: doesn't want a strict requirement
to review the original duty
... for interoperability we should have clear guidelines about
evaluation
<simonstey> not sat in one or all possible interpret.
michaelS: we could get different results if evaluators choose to include or ignore the duty with a consequence
LindaB: I feel a duty must be satisfied. In
Sabrina's example the duty had to be changed by a human
... so the original duty no longer worked
<ivan> https://www.w3.org/2017/Process-20170301/#revised-cr
ivan: the process document says - if its non
editorial we have to make a revised CR
... we might be able to do it without a call, but we will certainly have
to go through the process
... so its very urgent
<renato> https://github.com/w3c/poe/projects/2
ivan: we also still have 10 open issues
... if we go back we need to make sure these are resolved
renato: could the solution be - a small narrative change - with an editorial note to say you may have to look at impossible duties and get them relaxed
michaelS: where a duty cannot be fulfilled (e.g. the out of time one) - wants to repeat the original duty without a constraint if constraint means it can't be fulfilled
<simonstey> imo yes
michaelS: ideally people should consider what they are asking for when they set up a duty and if they expire - then plan for what happens then
renato: so this isn't just an editorial change
ivan: we'd need to extend the CR time
... and the 10 open issues must be closed
sabrina: we still need to agree.
... clarifies
... doesn't like making the duty part of the consequence
... I think if you have a duty then you must just fulfill it - a
consequence is more informational
benws: isn't it wise to say that this is implementational via the black box
<Sabrina> i agree with Ben that pushing the Obligation into the Consequence is a hack
<michaelS> Sabrina: again: how should issue https://github.com/w3c/poe/issues/275 raised by simonstey be resolved?
<simonstey> narrative as renato proposed
<simonstey> making people aware that they have to be aware that the original duty might need to be updated
<michaelS> you add a narrative to a Rule with a possibly unresolvable base Duty?
<simonstey> to keep it satisfiable
<renato> Proposal: Update the ODRL IM to state that to satisfy the duty (that triggers a consequence) MAY require implemenations to relax constraints (as an editorial change)
<michaelS> Who has to update: the maker of the Policy of its receiver?
<simonstey> theres no issue with duties having no refinements at all
<renato> Proposal: Update the ODRL IM to state that to fulfill the duty (that triggers a consequence) MAY require implementations to relax constraints (as an editorial change)
<renato> +1
<benws> +1
<Sabrina> +1
<ivan> +1
<michaelS> -0.5
<simonstey> +1
<benws> Victor - will you vote/
<benws> ?
<LindaB> +1
<victor> +1
ivan: further pleading about closing open
issues
... please
... goes back to editorial draft. make new section with changes since CR
[End of minutes]