Stream: IG creation
Topic: pb with fhirVersion 5.0.0-snapshot1
François Macary (Jan 28 2022 at 15:09):
I have a lot of errors and broken links when building the fhir-order-catalog IG in Continuous Integration, based on FHIR 5.0.0-snapshot1. It seems that the errors are triggered by this first one:
ImplementationGuide.fhirVersion[0] : error : The value provided ('5.0.0-snapshot1') is not in the value set 'FHIRVersion' (http://hl7.org/fhir/ValueSet/FHIR-version|4.6.0), and a code is required from this value set)
Is there a discrepancy of files somewhere?
Lloyd McKenzie (Jan 28 2022 at 15:12):
Yes. There's a difference in capitalization between what's required at different layers in the code.
François Macary (Jan 28 2022 at 15:18):
You mean the code needs corrections. Is it the code of the IG publisher or something wrong with files of the IG itself?
Lloyd McKenzie (Jan 28 2022 at 15:39):
I don't know what the 'right' answer is :)
Grahame Grieve (Jan 28 2022 at 18:34):
what's this about case?
Lloyd McKenzie (Jan 28 2022 at 18:57):
The code expects the suffixes on versions to be lowercase, but the enumeration expects them to be upper-case
Grahame Grieve (Jan 28 2022 at 19:01):
but the enumeration expects them to be upper-case
where?
Lloyd McKenzie (Jan 28 2022 at 19:03):
Schema? Not sure
Grahame Grieve (Jan 28 2022 at 19:04):
well, I can't find that.
Grahame Grieve (Jan 28 2022 at 20:32):
and I can't reproduce the problem either
Lloyd McKenzie (Jan 28 2022 at 21:07):
I ran into an issue where my Eclipse kindling build choked because the version it received was upper-case and the version it wanted was lower-case. I guessed that was the same thing happening here, though it might not be.
Grahame Grieve (Jan 28 2022 at 22:30):
shouldn't happen at all, but it's certainly not what's happening here
François Macary (Feb 01 2022 at 11:27):
Then do you have a clue of what is happening? The QA is available on the continuous integration:
https://build.fhir.org/ig/HL7/fhir-order-catalog/branches/master/qa.html
Grahame Grieve (Feb 03 2022 at 23:13):
I still can't reproduce that...
Grahame Grieve (Feb 03 2022 at 23:13):
and I don't have any ideas. what version are you using?
François Macary (Feb 04 2022 at 09:45):
Thank you @Grahame Grieve . @Rob Hausam and I have migrated the IG from 4.0.1 to 5.0.0-snapshot1. It is one of the couple IGs using this version in the CI, and it's there on Github: https://github.com/HL7/fhir-order-catalog/tree/master
What I don't understand is why our QA shows Build errors complaining about 4.6.0 and 4.0.1, which we are not using at all.
Grahame Grieve (Feb 04 2022 at 19:02):
well, it shouldn't. what version of the IG publisher are you using?
François Macary (Feb 09 2022 at 07:54):
@Grahame Grieve
It is the continuous integration, which produces this QA. Therefore the up to date version of the publisher, I imagine.
Just rebuilt today. The QA displays: IG Publisher Version: v1.1.99.
But still the same 83 build errors. Same kind as the one below:
MessageDefinition.status error The reference http://hl7.org/fhir/ValueSet/publication-status|4.0.1 could not be resolved
@Rob Hausam , I checked again ig.ini, package-list.json and input/ig-order-catalog.xml and I don't see any discrepancy. Could you do one more check on your side, please?
Rob Hausam (Feb 09 2022 at 10:44):
Yes, I'll take another look at that, @François Macary, and will see if I'm seeing the same thing ('llI try to do that today after the IPS call).
François Macary (Feb 10 2022 at 08:37):
@Rob Hausam did you have a chance to check? Could this strange mixing of FHIR versions searched by the publisher caused by some cache of packages somewhere? Given this IG started on 4.0.1, and then moved (unduly) to 4.6.0, and last to 5.0.0-snapshot1.
Rob Hausam (Feb 10 2022 at 16:40):
I haven't yet, @François Macary. If you haven't already, I agree that I would make sure that everything is deleted in your local cache - to be sure that this isn't coming from there. I will try to find a moment to look at it today. I guess we don't have a Catalog call tomorrow, but I'll follow up if there's anything that I can find.
François Macary (Feb 10 2022 at 16:51):
I have done that on my local repository: Removal of all the cache of packages. Sadly this did not bring any change to the errors on my local build. I have exactly the same QA as on the Continuous Integration.
François Macary (Feb 16 2022 at 14:02):
@Rob Hausam anything detected on your side?
Rob Hausam (Feb 16 2022 at 16:31):
I'm taking a look at the errors - most of them seem to be on a few of the extensions and CodeableReference. I don't have any key insights or solutions at the moment, but I'll see if I can dig in any further.
Grahame Grieve (Feb 16 2022 at 16:34):
the errors are in R5 core package
Rob Hausam (Feb 16 2022 at 16:37):
Do we have a solution or workaround?
Grahame Grieve (Feb 16 2022 at 16:39):
not at the moment.
François Macary (Feb 17 2022 at 09:46):
Will there be one before May? In other words, is a ballot of the IG in May still a reacheable goal?
Grahame Grieve (Feb 17 2022 at 11:17):
yes
Jens Villadsen (Mar 17 2022 at 10:24):
We also see a ton of errors on our IG at https://build.fhir.org/ig/hl7-eu/gravitate-health/qa.html - while some of them definitely are related to our own profiling and samples, I think that some of them are also related to internal stuff in the 5.0.0-snapshot1
Grahame Grieve (Mar 17 2022 at 12:43):
fix due out for that in the next 42 hours
Grahame Grieve (Mar 17 2022 at 12:43):
24 hours
François Macary (Mar 17 2022 at 19:18):
Yes, I see that gravitate-health IG and order-catalog IG have the same kind of QA, with, in addition to the errors already mentioned in this topic, hundreds of thousands of broken links.
François Macary (Mar 18 2022 at 15:58):
Love version 1.113! Thank you @Grahame Grieve
Lots of issues fixed.
Last updated: Apr 12 2022 at 19:14 UTC
