Stream: implementers
Topic: Syntax for documenting chaining capabilities
Lloyd McKenzie (Dec 06 2021 at 21:53):
In SearchParameter, there's an element called 'chain' of type string. But we don't actually document how it's used. Specifically:
- do you list just the next level (with the presumption that people can nest as deeply as they like) or do you list all specific chain combinations to the depth you support them (meaning you might support chaining to depth 1 and a couple of depth 2 chains, but not all of them)
- what is the syntax to describe what chains you support. Do you include the current search parameter name, or just what's below it? Do you lead with a "." or not?
See tracker item here: FHIR#32531
Grahame Grieve (Dec 06 2021 at 23:02):
do you list just the next level (with the presumption that people can nest as deeply as they like)
That's what test.fhir.org does
Grahame Grieve (Dec 06 2021 at 23:03):
do you list all specific chain combinations to the depth you support them
That's a permutational explosion that is not closed on test.fhir.org
Grahame Grieve (Dec 20 2021 at 20:21):
@James Agnew @Alexander Zautke @Paul Church @nicola (RIO/SS) @Lee Surprenant @Peter Jordan @Michael Lawley @Brian Postlethwaite @Grahame Grieve - can you comment?
nicola (RIO/SS) (Dec 20 2021 at 20:24):
It's very inefficient, but with aidbox you have unlimited "chains". The common problem with chain, that there is no way to put two or more criteria on same chain( join)
nicola (RIO/SS) (Dec 20 2021 at 20:24):
We even think to introduce "_join" parameter for that
nicola (RIO/SS) (Dec 20 2021 at 20:25):
But for security and performance reasons from other side we often need to restrict, which chains are allowed
nicola (RIO/SS) (Dec 20 2021 at 20:27):
I.e. from api point of view it would be great to document this restrictions
Brian Postlethwaite (Dec 20 2021 at 21:43):
Ours is open like test.fhir.org. So would be impractical to express.
And still yet to implement chaining over a canonical reference.
Lloyd McKenzie (Jan 10 2022 at 21:43):
Didn't get a whole lot of responses, and I'm nervous that those solicited are all those who have 'generic' FHIR servers - which will typically always support unlimited chaining. What about those who have legacy back-ends who only support very specific types of chains and don't necessarily let you chain through to everything and might well want to enumerate? How do we differentiate the declarations of the generic folk (for whom listing a search parameter automatically means "and anything you want beneath it" from the simpler systems for whom it means "and this alone, with no further nesting".
Last updated: Apr 12 2022 at 19:14 UTC
