Should DID syntax allow an empty "method-specific-id"?

[brentzundel]

@peacekeeper moved from CCG (w3c-ccg/did-spec#198)

[peacekeeper]

I've been thinking about this for a while and I now have a slight preference, but I'd like to share some considerations first.

What is the DID subject in the case of an empty method-specific identifier? The answer that was given in the past was that the DID identifies the DID method or DID registry itself. But then, what does it mean...

• to prove control of this DID in a DID Auth interaction?
• to discover means for trustable interaction (services and keys) with this DID subject?
• to have a VC that is issued by this DID?
• to execute the CRUD operations on this DID?

I admit that some of the above sounds intriguing and could make sense. But on the other hand, an empty method-specific identifier feels a bit like a special case that is quite different from what DIDs have been designed for.

What I'd like to avoid is allowing this merely for being able to discover metadata about a DID resolver, or for some kind of general-purpose query mechanism to the DID registry. Those things should be handled differently, not by piggybacking on DID syntax and DID resolution. Instead, the question we need to answer here is "does it make sense for the DID registry to be considered a DID subject itself".

Thoughts?

[peacekeeper]

Since there has been no activity on this issue, my (slight) preference would be to NOT allow an empty method-specific-id. I don't feel strongly about this and could be convinced otherwise, but my sense right now is that the above questions have the potential to lead to ambiguity and confusion (i.e. what is the DID subject/controller, what is the meaning of the CRUD operations, etc.).

So unless anyone feels differently, I propose to raise a PR to change the ABNF to NOT allow an empty method-specific-id.

[peacekeeper]

@talltree ^

[msporny]

does it make sense for the DID registry to be considered a DID subject itself

No, the only use case for it is weak and can be solved through a variety of other mechanisms like a regex or matching against "did:v1" or a variety of other things that are perfectly fine and don't need to comply w/ DID syntax.

[peacekeeper]

Thanks @msporny I agree. Others please add thoughts. If there is no opinion to the contrary, we can create a PR to change the ABNF to no longer allow an empty method-specific-id.

[dlongley]

Agree with @msporny.

[kdenhartog]

PR looks good and I agree with the consensus that's been reached.

[peacekeeper]

This has been addressed by #216. Closing.