Stream: implementers
Topic: Document summary
Elliot Silver (Jan 18 2022 at 22:36):
If I have document bundles stored at /Bundle, what should I expect to see if I issue a query with _summary=text? It seems to me, by the spec, I should only get back id, meta, and type, which isn't very useful. (Bundle doesn't seem to have a text element, since it's not a DomainResource.)
Elliot Silver (Jan 18 2022 at 22:41):
If I issue a query with _summary=true, what is the expectation for what gets included from the bundled resources? (entry.resource is a summary element.) Do I get the whole resource, or only summary elements of it, or something else?
Elliot Silver (Jan 18 2022 at 22:45):
Since, every element in Bundle is marked Summary, is there a difference between a _summary=true query and a non-summary one?
Craig McClendon (Jan 18 2022 at 22:50):
I would expect /Bundle/123?_summary=true to return the entire Bundle as-is with no attempt to summarize the entry resources.
I would expect /Bundle/123?_summary=text to return just the 3 attributes you mentioned. I don't know that there's anything stopping you from adding extra attributes to the summary text, though.
edit to add: this is just from my reading of the spec. I don't have any particular experience with FHIR document paradigms, so could be other folks have more to add.
Elliot Silver (Jan 18 2022 at 23:03):
That was my reading of the spec too. I'm just not sure it's particularly helpful behaviour.
John Moehrke (Jan 18 2022 at 23:42):
further... the definition of _summary https://www.hl7.org/fhir/search.html#summary includes
Servers are not obliged to return just a summary as requested.
so, it is always completely at the whim of the server.
Elliot Silver (Jan 18 2022 at 23:44):
I know, YMMV. But I was wondering what a client should expect.
John Moehrke (Jan 18 2022 at 23:48):
I think any client using _summary should have very low expectations. The only _summary function that likely has worthwhile expectation is count.
Lloyd McKenzie (Jan 19 2022 at 05:36):
Everything in Bundle is in summary except extensions, so I'd expect extensions stripped out but not anything else. Whether that's useful or not, I think it's too late to change...
Gino Canessa (Jan 19 2022 at 06:03):
I believe there was a ticket or two to clarify around this… I think the resolution was that once an element was included, then everything below it was included. But, I would suggest searching for the ticket.
Elliot Silver (Jan 19 2022 at 17:29):
Is it too late? I know Bundle is normative, but summary behaviour is largely at the discretion of the implementation, so changing the expected behaviour wouldn't cause existing implementations to be non-conformant.
Gino Canessa (Jan 19 2022 at 17:32):
Adding clarifying text is something that can always happen. Related to _summary behavior, there were two related tickets resolved (voted on) during FHIR-I sessions today: FHIR-33002 and FHIR-33148 .
Lloyd McKenzie (Jan 20 2022 at 02:35):
Summary behavior is not at the discretion of the implementation. If you have an element and the requester is allowed to see it and they as for _summary=true and you don't provide it, you're non-conformant.
Lloyd McKenzie (Jan 20 2022 at 02:36):
Flagging a parent as _summary doesn't automatically mark all children as _summary. The decision to mark everything in the Bundle as 'isSummary=true' was a deliberate action involving setting the flag on all rows in the resource. I can't explain why - @Grahame Grieve might be able to when he returns. But it's not something that could have happened by accident.
Grahame Grieve (Jan 27 2022 at 05:05):
not sure I remember
Grahame Grieve (Jan 27 2022 at 05:05):
maybe it was an expedience thing, since bundle has to be exempt from summary when it's the result of a search
Last updated: Apr 12 2022 at 19:14 UTC
