FHIR Chat · JSON duplicate properties

Stream: conformance

Topic: JSON duplicate properties

Michael Lawley (Feb 07 2018 at 23:27):

JSON itself does not forbid duplicate properties, although the spec does observe that they hurt interoperability since various libraries behave differently.
The FHIR spec for application/fhir+json describes how to generate "valid JSON" and the process will not produce duplicate properties.

Are duplicate properties explicitly invalid and thus should a server always report an error when one is encountered?

Lloyd McKenzie (Feb 07 2018 at 23:46):

I would say they're always invalid. If it's a repeating property, all repetitions should appear in the array of a single property. If you'd like to submit a change request for us to make this super explicit, that's not a bad idea.

Michael Lawley (Feb 08 2018 at 02:44):

Done - GF#15290

Grahame Grieve (Feb 08 2018 at 02:51):

yes always invalid.

Michael Lawley (Feb 08 2018 at 05:14):

@James Agnew Looks like HAPI's use of GSON means it silently ignores this - I've filed a bug :)

Lloyd McKenzie (Feb 08 2018 at 05:21):

Technically, servers don't have to reject invalid instances, but agree that this would be a good thing for HAPI to catch.

Jim Steel (Feb 08 2018 at 05:48):

or to be able to catch, at least

Ewout Kramer (Feb 15 2018 at 13:10):

@James Agnew Looks like HAPI's use of GSON means it silently ignores this - I've filed a bug :)

This is exactly what .NET's json parser (Newtonsoft) does - and there's no way to detect it. People have tried to file that as a bug, but the author is pretty stubborn about it. So -at least in .NET- I would not be able to enforce it :-(

Michael Lawley (Feb 17 2018 at 09:19):

I've seen similar threads discussing the issue wrt other JSON parsers too. This doesn't seem like a good situation. I'm wondering how this might interact with digital signatures and Canonical JSON