Hi Shane/Manu,
We also don't specify this chain-halting feature anywhere in the Syntax
Document or the Primer, do we?
I do not believe that we do. Not explicitly anyway.
Anyway...the key point is that we never intended to 'halt chaining';
there was no use-case, discussion or vote on that kind of feature.
@rel="" popped up only in the context of the discussion I've just
mentioned, and even then is was 'in passing'.
I guess that what I am trying to get at is there is no definition of
what should happen in the event that a rel="" is encountered. My
reading of the processing rules says that rel="" is not a valid CURIE,
so it is ignored. If everyone agrees with that, then I think we can
codify it in the section on CURIE processing and move on - e.g., "For
the avoidance of doubt, a CURIE that evaluates to "" is NOT a valid
CURIE." Or something like that. Then we are explicit and no one can
argue about it later.