See also: IRC log
benws_: Wasn't this what we had? Two ideas (1) The validator verifies the syntactic validity, i.e. compliance with the IM (2) The evaluator determines the active rules
michaelS: duties, remedies and consequences are not checked (see section 2.6.8)
victor: as a reminder, see the tables at https://docs.google.com/spreadsheets/d/1vej4EwEZjUM2yYJwBggJxHlSTCinyuMPCulKIeHNTA0/edit#gid=421997599
michaelS: if you have the permission you have to check if the rule is active, and then, after, the duties. (The constraints are already verified in the first stop)
<renato> https://w3c.github.io/poe/model/#rule-active
michaelS: we were unsure about that, happy to discuss with you now that
simonstey: indeed, you can change the order and verify the duties first. But perhaps you discover later that the constraints were not been valid.
benws_: I agree with everything, but perhaps these details are not necessary in the spec
lindab_: can you please provide an example?
benws_: let us consider a complex case with several duties and consequences. If I check first that the duties are not satisfied for a rule triggering others, then I do not need to evaluate the entire chain.
victor: verfiying if a duty is fulfilled is making an observation. Sometimes making this observation is not "free". In this case, for a particular implementor, the order of the verifications (constraints/duty vs duty/constraints) may have an impact. If we are neutral wrt, everybody will be happy
<simonstey> https://github.com/simonstey/ODRL-SHACL-Shapes
<simonstey> https://github.com/simonstey/ODRL-SHACL-Shapes/wiki/3.1.4-Agreement
simonstey: please see the wiki above to see
some validations
... the validator would have to be able to identify that invalid
examples listed in the wiki are actually invalid
<renato> ODRL Features: https://github.com/w3c/poe/blob/gh-pages/test/cr-exit.md
simonstey: being exhaustive with the shapes and digging in the text is tedious
<renato> we lost u ben!
<renato> call droped
benws_: I face the same situation with the evaluation
Ben, do you read us at least?
renato: I remind the deadline is September 11th
michaelS: must I implement a RDF shapes processor? or can I use other implementations leading to the same result?
simonstey: you are absolutely free
<renato> XML is informative
<simonstey> that's what we are trying to figure out .:)
victor: (1) can I ignore XML
others: yes
victor: (2) how shall we verify that implementations are conformant?
benws_: there must be a report (with a given format, like SHACL)
simonstey: we have not time for that
... there is no standard on how the implementation report should look
like
... fancy systems are probably intended for specs with hundreds of
implementors
benws_: who else will help to extract contraints?
victor: +1
michaelS: +1
... how is materialized the help?
simonstey: text describing constraints (like cardinality etc.)
victor: Ideally I would be happy to have an example and a counterexample
benws_: the examples in the IM must be updated with the new refinements, I hope that by Monday they are all correct and agreed
renato: I will work in the evaluator, I hope to have something by Monday
simonstey: examples 1 and 2 for the constraints are faulty, the id is wrong for the json-ld
victor: not having yet the full set of shapes and validations; and with the evaluation part also under construction, please indulge if the implementation I present next week does not fully work
michaelS: work split: victor will look from from policy to action, whereas michael constraints and rules