Closed simonstey closed 7 years ago
I have made the suggested changes. Please see the diffs here: https://github.com/w3c/poe/commit/b4c6fca8cced6b13c34a15e227f32e065ff1f6da
Comments:
1) We use RFC2119 for consistency 2) Example 21 is an Offer and should not define an Assignee 3) Examples aid in better understanding the specification. If there are ones that are incorrect or could be updated, then please indicate. 4) We plan to have the ODRL Best Practices NOTE to include more examples.
- We use RFC2119 for consistency
AFAIK (and others, e.g., @philarcher1 or @iherman please correct me if I'm wrong) using a RFC2119 key word like MUST NOT for a statement like A Prohibition indicates that the Policy expresses an Action that MUST NOT be performed,
would require any implementation using ODRL to be actually able to verify whether prohibited action is indeed not performed in order to conform to the standard. Can we require that?
- Example 21 is an Offer and should not define an Assignee
yes, but how does this align with a Duty states that a certain Action MUST be executed by the Party with the Role Assignee
? Does that mean you cannot define duties for any policy type that does not require an assignee? what happens if there's a duty defined for a permission that is missing an assignee?
- Examples aid in better understanding the specification. If there are ones that are incorrect or could be updated, then please indicate.
I'm def. not against having examples in the spec.. I just think that having informal ones like Alice must pay 5 Euros in order to get the Permission to play abc.mp3.
at the very beginning is not necessary & confusing, esp. when proper examples are provided later on anyway.
- We plan to have the ODRL Best Practices NOTE to include more examples.
perfect! again, I'm not against the actual examples (i.e., the JSON-LD ones)
1) Don't forget that ODRL is an expression language, not an enforcement language :=) 2) You can always express a Duty, but who does it depends on other factors, eg an Offer being "accepted" an turned into an "Agreement"
Note on a detail:
Removed the "digital" only wording...
what? how's the context of an ODRL policy defined? how does one execute the use? the use of what? Given that ODRL defines both
odrl:use
andodrl:execute
, this might cause confusion.Isn't that domain specific? Why do we need to make those statements?
that's not only limiting the applicability/usefulness of ODRL in general, but also not reflected in the examples of the spec.
phrasing; do we need that part at all? -> remove
do we need to mention profiles here? -> remove
too much prose.. also s/entities/classes, s/subtypes/subclasses
what Rules? they all have the same relationships to other concepts? why do we need to make this statement?
up to this point the reader does not know what a Party and/or Action is. What about assets? What if no party is defined? This list also suggests that a duty is a standalone concept, i.e., doesn't need to have a sorrounding permission to be applied to (which is wrong). I would also not use RFC2119 terms for non-implementation specific things.
what's the purpose of this statement? -> remove
we don't need to sell ODRL to anyone + inappropriate for a W3C recommendation
again, executed as in
odrl:execute
? what's a related asset?which means? does it limit the validity of the rule it is defined for or are two rules one w/ and one w/out a constraint actually equivalent, but one is more precise than the other?
to do what?
so..? do we need that paragraph at all? despite being hard to understand, we do explain all the concepts later on anyway.
wait what? why is this similar to permissions? for what permission? what if there's no party with role assignee defined (like in Example 21)? What's the intended semantics of an invalid permission (or rule in general)? remove example
so they are not used in the same way.. also (imho) misleading use of RFC2119 keywords; how does one refer to a duty? the Action -> what action? forbidden == prohibited? remove example
so why do it sloppily here?
My suggestion:
fwiw, I very much like SHACL's intro section