FHIR Chat · Unknown search parameter repetition · implementers

Stream: implementers

Topic: Unknown search parameter repetition


view this post on Zulip Alexander Henket (Nov 14 2019 at 11:55):

Suppose I have an IG that says to use param X once and with 1 value:

  • Is there a default expectation for a compliant server on how to deal with repetitions of param X?
    ** Ignore (or Support if it chooses to do so),
    ** Refuse

  • Is there a default expectation for a compliant server on how to deal with repetitions of the value of param X?
    ** Ignore (or Support if it chooses to do so),
    ** Refuse

view this post on Zulip Grahame Grieve (Nov 14 2019 at 12:08):

unknown parameters should be ignored?

view this post on Zulip Alexander Henket (Nov 14 2019 at 12:15):

Is the second repetition of a known parameter considered "unknown"?
Is the second value of a known parameter considered "unknown"?

view this post on Zulip Grahame Grieve (Nov 14 2019 at 12:20):

no it's still a known parameter. What it means depends on the parmaeter

view this post on Zulip Alexander Henket (Nov 14 2019 at 12:27):

Sure, but the second occurrence or value was not specified so it might not be catered for. Would the server logic then be expected to ignore that extra occurrence or value (Prefer: handling=lenient), or could it refuse to answer (Prefer: handling=strict). Or should the server logic always be prepared for repetitions of value, but not necessarily for repetitions of the parameters?

view this post on Zulip Grahame Grieve (Nov 14 2019 at 12:32):

it should be prepared for both

view this post on Zulip Alexander Henket (Nov 14 2019 at 12:34):

So that means an IG could not meaningfully constrain repetitions of either param or its value? Or by preparation you mean, being able to detect them and not simply skip them?

view this post on Zulip René Spronk (Nov 15 2019 at 04:52):

If the server had its own business logic which constrains the use of a particular parameter to one occurence, I'd definitely expect the server to create an error. We want P=a&P=b to be isomorphic with P=b&P=a if P is a known parameter. The problem is of course that this constraint use case has never been considered, and as such Alexander may just have to put forward an issue with a proposal as to what wording should be added to the standard.


Last updated: Apr 12 2022 at 19:14 UTC