Stream: IG creation
Topic: Missing ValueSets using -tx
Jim Steel (Dec 04 2019 at 01:47):
I'm trying to (locally) build an IG, and I'm getting a lot of reports like No value set found at Medication.form (url = 'https://healthterminologies.gov.au/fhir/ValueSet/medication-form-1'), despite the ValueSet being present on the terminology server that I'm specifying using -tx. How is the publisher determining whether a tx server has a ValueSet? Not using /ValueSet?url=XX or /ValueSet/$expand?url=XX ?
Grahame Grieve (Dec 04 2019 at 01:51):
I don't think that error comes from the terminology server
Jim Steel (Dec 04 2019 at 01:51):
No, I had assumed it was the publisher reporting a problem with the terminology server
Jim Steel (Dec 04 2019 at 03:32):
But I'm not sure what that might be
Jim Steel (Dec 05 2019 at 00:12):
Is the IG publisher tool looking for something about ValueSets in the terminology capabilities statement?
Grahame Grieve (Dec 05 2019 at 00:18):
it's trying to render the profile. The profile refers to the value set, but the IG publisher has no knowledge of where to point the value set reference to
Jim Steel (Dec 05 2019 at 00:23):
Looking at my tx server logs, it also only seems to be performing CodeSystem/$validate-code checks. No ValueSet?url=or ValueSet/$validate-code (or a lot fewer than I would expect)
Grahame Grieve (Dec 05 2019 at 00:24):
did you clear the txCache before looking at your log?
Jim Steel (Dec 05 2019 at 00:28):
I'm running with -resetTx, does that do the same thing?
Grahame Grieve (Dec 05 2019 at 00:28):
y
Jim Steel (Dec 05 2019 at 00:28):
then yes
Jim Steel (Dec 05 2019 at 00:33):
It might be that some of the command-line "missing ValueSets" messages might be due to "missing" ValueSet urls in the ig.json (the ADHA ValueSet urls do resolve for a human, but not quite properly for a program that expects to resolve to JSON/XML)
Grahame Grieve (Dec 05 2019 at 00:34):
when you refer to a value set in a profile, you need to not only make it available via a terminology server, you also need to make it available as a published artifact, so that it can be rendered properly. That means, published in an IG that is registered as a dependency
Jim Steel (Dec 05 2019 at 00:34):
Yeah, that's very desirable. Not the case at present for a lot of these national terminology artifacts
Grahame Grieve (Dec 05 2019 at 00:35):
then you get that note, but it otherwise publishes
Jim Steel (Dec 05 2019 at 00:38):
Yep. I'll turn my attention to the validation errors. I'm still getting a lot of "ValueSet https://healthterminologies.gov.au/fhir/ValueSet/australian-medication-1 not found by validator" in the qa.html despite the ValueSet being in the tx server
Grahame Grieve (Dec 05 2019 at 01:05):
it's following the same logic - trying to find the value sets in the dependencies
Jim Steel (Dec 05 2019 at 01:16):
So there's no way to have them from outside a dependency? That's pretty limiting when using authoritative third-party terminology out of the control of the IG author
Grahame Grieve (Dec 05 2019 at 01:27):
I'll think about it. It's not a small upgrade...
Jim Steel (Dec 05 2019 at 01:27):
Sure
Jim Steel (Dec 05 2019 at 01:30):
I mean, we might have levers to make/help ADHA build a package as an additional delivery mechanism for their terminology resources, but I feel like it will become a really difficult additional sell when we're trying to convince other terminology sources that they should produce FHIR versions of their artifacts. Its hard enough convincing them to produce a CodeSystem resource
Grahame Grieve (Dec 05 2019 at 01:30):
CoeSystems we are much more flexible about
Jim Steel (Dec 05 2019 at 01:30):
Potato, potato
Grahame Grieve (Dec 05 2019 at 01:30):
but how would it work if you don't have FHIR Value set?
Jim Steel (Dec 05 2019 at 01:30):
;)
Jim Steel (Dec 05 2019 at 01:31):
As in, nothing either at the resolved URL, nothing in the IG itself, nothing in the tx server?
Grahame Grieve (Dec 05 2019 at 01:32):
well, if you have it in the tx server, you have it as a fhir resource...
Jim Steel (Dec 05 2019 at 01:33):
Yes, that's my current situation
Grahame Grieve (Dec 05 2019 at 01:35):
with regard to validation... I can re-org the code internally to deal with only having the value set on the tx server. But for publication... I need to know the html location for human consumption
Jim Steel (Dec 05 2019 at 01:35):
Yep, that seems reasonable
Jim Steel (Dec 05 2019 at 01:36):
As a next step
Grahame Grieve (Dec 05 2019 at 01:36):
so we'd somehow need to figure out how the renderer figures out the html location for a a value set it holds. It might not be the literal URL...
Jim Steel (Dec 05 2019 at 01:39):
On a related note, just while I'm debugging stuff, does the resolution of these terminology resources by URL follow 301 redirects?
Grahame Grieve (Dec 05 2019 at 01:42):
I don't know. It would depend on how the http client is configured
Grahame Grieve (Dec 05 2019 at 01:42):
and I don't remember
Jim Steel (Dec 05 2019 at 01:43):
I've just been looking at the healthterminologies ValueSet URLs that the publisher is complaining about, and they do actually resolve, but only by following the 301s
Grahame Grieve (Dec 05 2019 at 01:44):
well, when we try to resolve them, I'll keep that in mind
Michael Lawley (Dec 05 2019 at 04:00):
Would it work to have the ValueSet in the tx server point at a location for human consumption (or something in the IG itself)?
Grahame Grieve (Dec 05 2019 at 04:52):
that could work (Won't now)/ But how would the IG publisher know?
Last updated: Apr 12 2022 at 19:14 UTC
