When building data intermediaries with FHIR, I've wished a clearly defined .identifier.type would exist for carrying the
.id a resource has in the originating server. I do create separate Provenance resources and tag the information there. But I would also use an identifier if there was a clearly documented way for doing that. To complement what I can do with .meta.source, where I can say where the data comes from.

Having a specified mechanism for adding resource id's as identifiers would help many implementations in deduplication efforts.

Take the IPS specification as an example. An implementation composing the IPS may fetch data from multiple sources. Various IPS bundles, with various versions and updates, can then be spread around and data from them may be extracted to various systems. It helps these systems quite a bit if the origin of the individual resources can be read from the resource.

I guess any well specified identifier would help in deduplicating information. However, since it seems there is no generally used identifier, the resource id on the originating server could be one, when used together with the information of the originating server (in the identifier, use the server in system and id as value). Another, and perhaps better, option is to use the full URL of the resource on the originating server as the value.

I did check the existing codes in the identifierType code system. There is the code RI for Resource Identifier, but that does not seem to be a good fit for a FHIR resource. None of the other codes seem like a good fit either.

So I'm proposing to add a new .type code.

code: FURL
display: FHIR resource URL
description: The full URL of the FHIR resource on an originating server. Preserving the URL as an identifier is useful especially for deduplication purposes when the resource is transmitted away from the server.

See some discussion in Zulip.