See also: IRC log
<michaelS> scribe: CarolineB
<michaelS> scribenick: CarolineB
<renato> https://www.w3.org/2017/01/23-poe-minutes
<phila> [NOTUC]
minutes approved
<phila> (No objection to unanimous consensus)
<renato> https://github.com/w3c/poe/issues
renato: from issues list, got
thorugh many last week, couple waiting for comment.
... semantic details being discussed
... json ld section needs work
<simonstey> +q
benws2: complex constraints. Simon and Michael were going to give examples to show pitfalls
<renato> isNecessaryFor - the left operand Constraints must be satisfied before processing the left operand Constraint which then must be satisfied
simonstey: main problem is that isNecessary means that temporal dependencies can be imposed between constraints
<simonstey> c1 isNecessaryFor c2, c2 isNF c3, c3 isNF c1
simonstey: adn this leads to deadlock - as example above
<simonstey> c1 isNF c2, c3 isNF c2, c1 xor c3
<simonstey> c4 isNF c1 -> c4 not resolveable
<michaelS> q
simonstey: I think we should
avoid combining constraints so that we avoid huge chains
... lets simply say before or after date/times
michaelS: we need to think of
worst cases. So if checking takes too long we could set a
timeout period for checking
... or a default value for what happens if you get a
timeout
renato: Remove all extended constraint stuff?
simonstey: just leave and or xor
etc. remove IsNecessary concept for now
... we need to leave constraint on constraint, not remove
it
... just still need to think about the requirement. Temporal
constraint is too difficult
benws2: we d need to be able to
check amount and unit
... so we need to keep case where there is more than one
predicate. Is it simply temporal that is problematic?
... are there complex constraints we can retain?
renato: haven't yet found a
solution to cover all cases despite talking a lot
... but if we don't have it at all in the draft we are saying
we can't handle constraint on constraint at all
<benws2> +1 to expression
simonstey: if we keep it, I want
to go over edge cases and explain the problems
... we must avoid must be satisfied *before* processing. to
avoid deadlocks
michaelS: so should we include sequence for processing?
renato: is necessary for gives us that sequence
brian_ulicny: owl and sparql use
if then - can we do the same and only evaluate all constraints
when they are all satisfiable?
... exampe of event + 30 inutes relies on representing the
extra 30 mins
simonstey: All Time a possible solution?
<simonstey> owl
owl time that is
<simonstey> https://www.w3.org/TR/owl-time/
simonstey: you can't always check everything of ocurse
<phila> phila: if it matters, OWL time gets an update on Thursday https://www.w3.org/TR/2017/WD-owl-time-20170202/
<benws2> :)
renato: we will need to revisit this section and add more narrative to IsNecessaryFor (isNF)
simonstey: some editorial things
need fixing.
... current information model still contains legacy sections
(from xml) that are confusing; e.g. 3.4
... adn some parts too implementation specific; e.g. inherit
condition
... please would everyone go over, check these things and
suggest changes w eneed
benws2: there is some ambiguity around prohibitions and permissions
simonstey: shouldn't have a policy with only prohibitions
benws2: is a prohibition at same level as permission or should it only qualify a permission?
renato: originally anything
permitted must be in the expression all else assumed ot be
prohibited
... yes, we should probably revisit for clarity
michaelS: the starting point of nothing is permitted came from discussion with lawyers. It's a legal pov
simonstey: suggests this could be too restrictive for ODRL
<smyles> http://dev.iptc.org/RIghtsML-Processing-Model
smyles: rightsML processing model
is defined 1. first look at whats permitted then look at
ODRL.
... reason we defined it that way since news is handled that
way now ex ODRL
renato: we are reviewing the prohiition section
<simonstey> +q
renato: q-
... we need to create a JsonLD context
<michaelS> regret not joining next call