Re: Rules for matrix parameter vs. resolver option (was: Re: Proposed work item: did:key DID Method)

+1 to not overusing matrix parameters.

There is a green note box at the end of
https://w3c.github.io/did-core/#generic-did-parameter-names with some
language about when to use matrix parameters and when to use resolver
options.

It currently says:
- Use matrix parameters to "specify what is being identified"
- Use resolver options to "control how that is dereferenced"

Your language is also good I think, maybe we can fold that in somehow.

To me, the important distinction has been whether or not you need a DID
URL that can be referenced, linked to, used in an RDF graph, a database,
etc.
This is quite similar to the question when do you put something into an
HTTP URL, and when do you use HTTP headers.

We have this on the agenda for the WG call today.

Markus

On 11/25/19 11:20 PM, Manu Sporny wrote:
> On 11/24/19 10:28 PM, Markus Sabadello wrote:
>> It may actually be better to do something like this with a resolver 
>> input option, rather than a DID URL matrix parameter.
> I'd personally prefer it as a resolver input option. Again, we can stuff
> a ton of functionality into matrix parameters, but just because we can
> doesn't mean we should.
>
> What we don't have is a clear set of rules for when something should be
> a matrix parameter and when something should be a resolver input option.
>
> Attempting a first cut at those rules here:
>
> Matrix parameters are used when you need a mechanism to store a DID with
> information that's critical to the intended usage of the DID. For
> example, writing a DID to a database where you want to specify that you
> want to use a very specific version of a DID Document associated with a DID.
>
> Resolver input options are used when what is critical to the usage of a
> DID Document combination cannot be predetermined. For example, whether
> an entity wants the resolver to resolve the public keys in the DID
> Document in JWK format or in native format.
>
> Don't know if others agree with those rules, or have a suggestion on a
> different set of rules when one should pick between using a matrix
> parameter or a resolver option.
>
> -- manu
>

Received on Tuesday, 26 November 2019 09:21:17 UTC