21P/ Universal Resolver Driver Policy Discussion

Universal Resolver Driver Policy Discussion

Thursday 21P

Convener: Bernhard Fuchs, Markus Sabadello

Notes-taker(s): Markus Sabadello

Discussion notes, key understandings, outstanding questions, observations, and, if appropriate to this discussion: action items, next steps: 

Currently, instances of the Universal Resolver is hosted by DIF, IBM, and other companies. Danube Tech has been maintaining the project.

The project has some guidelines for contributing new DID method drivers:
https://github.com/decentralized-identity/universal-resolver/blob/master/docs/driver-development.md

We have some ongoing questions on policies for Universal Resolver drivers.

Proposal: We should require contact data for maintainers of drivers (could be email address or any other type of contact data).

Another challenge is that there may be multiple projects claiming the same DID method name. How to decide which DID method driver to include in the Universal Resolver?

Proposal: Driver implementers must get their DID method registered first in the W3C DID method registry, then they can contribute a Universal Resolver driver (this avoids ambiguities)

DID test suite: https://github.com/w3c/did-test-suite

DID test suite is not for runtime, but the Universal Resolver could do a few simple checks on a driver's responses. But there's also a philosophical question: Should the Universal Resolver be "allowed" to check and potentially transform driver responses, or should it just "pass through" everything that comes from a driver?

What should be the Universal Resolver policy once DID Core reaches v1.0? Remove all non-compliant drivers, or automatic translation in a transition period.

Some discussion also about the Universal Registrar, which is a similar project, but designed for DID write operations rather than DID resolve.