Mike Ralphson

> Different processors could accomplish this in different ways, and Canonical Form does not guarantee that the output will always be exactly the same, regardless of which processor you use....

On a first reading, this looks to be a fantastic set of cleanups, improvements and fixes. One thing that might still be outstanding is to clarify the schema name resolution...

Also, if the `discriminator`is used in a pure JSON Schema resource, it is likely that there will be a `$defs` section, not a `components/schemas` section. Do we allow those shortcuts?

I think I agree with the thrust of what you say there @jdesrosiers > Therefore, the "schema name" addressing feature of discriminator can't be used in a pure JSON Schema...

Very open to dealing with additional issues outside of this PR, no need for `mapping` considerations to block this.

I think more generically `service-desc` might suit the bill. See https://tools.ietf.org/html/rfc8631 See #1511 #724 #734 for previous discussions including `describedBy`, and #110 for a proposed media-type for OAS documents.

When #110 (@darrelmiller) is resolved we would definitely look at an updated version of #734 wording, either in the spec. or in a guidance / technical note document.

I recall @darrelmiller had some objection to using a `.well-known` URI for this case - could we document it here? It may have been that an OpenAPI document doesn't meet...

Thanks. Do you have time to PR a change? We don't have an open 3.0.4 branch at the moment, but as this is the first time this has been noticed,...

@deus-x-machina that sounds like an issue with the specific tool you're using - not the spec.