Open megoth opened 4 years ago
You mean triples like <#owner> rdf:type acl:Authorization
? Yes, definitely required to make the semantics work. So that's a double bug then indeed, good catch! :)
Actually a triple bug, Solid UI + NSS + test-suite. :)
Ok, I'll create separate issues for NSS and test-suite for this. I'll also try to find the appropriate sections in the Web Access Control specification.
I found the issue "Is acl:Authorization mandatory?" in the WAC spec, where the question is discussed as part of the new specification.
I'll keep an eye out for those discussions, but until there's a conclusion I'll wait to create issues on NSS and the test-suite.
That being said, for the sake of semantics I suggest that we make sure that acl:Authorization
is added as part of the code we touch in Solid UI and other places. (I've made sure to fix this in in adoptACLDefault
in a PR that is coming very soon.)
But meanwhile, we can still fix the fact that the copy doesn't copy it, then, so this issue is still a bug.
I'm writing tests for the acl module, and noticed that
adoptACLDefault
does not copy the triples containingacl:Authorization
into the new graph it builds. It seems that NSS does not care about this triples though, which is why the bug has gone unnoticed.I'm creating this issue to verify my suspicion with others. If it is the case that NSS should require
acl:Authorization
triples, I'll create a bug there as well.