Stream: IG creation
Topic: search parameter warning
Eric Haas (Jul 01 2020 at 23:07):
I am getting this warning on several search parameters...
SearchParameter/us-core-careplan-date: SearchParameter warning SEARCHPARAMETER_EXP_WRONG
if this means the expression is wrong that is curious since the expression are identical to the corresponding base searchparameter:
for example: CarePlan patient
ig version:
...
"expression" : "CarePlan.subject.where(resolve() is Patient)",
"xpath" : "f:CarePlan/f:subject",
...
corresponds to the http://hl7.org/fhir/SearchParameter/clinical-patient
Eric Haas (Jul 01 2020 at 23:09):
these only seem to occur with the common SearchParameters like clinical patient
Grahame Grieve (Jul 02 2020 at 01:48):
@Eric Haas I found out why you were getting "SEARCHPARAMETER_EXP_WRONG". here's the errors you should be getting:
Grahame Grieve (Jul 02 2020 at 02:21):
actually, there are none now that I improved the checking.
Grahame Grieve (Jul 02 2020 at 02:23):
wrong:
Grahame Grieve (Jul 02 2020 at 02:25):
- The expression "DocumentReference.id" is not compatible with the expression "Resource.id" in the derivedFrom SearchParameter http://hl7.org/fhir/SearchParameter/Resource-id, and this likely indicates that the derivation relationship is not valid
- The expression "Location.name" is not compatible with the expression "Location.name | Location.alias" in the derivedFrom SearchParameter http://hl7.org/fhir/SearchParameter/Location-name, and this likely indicates that the derivation relationship is not valid
- The expression "Organization.name" is not compatible with the expression "Organization.name | Organization.alias" in the derivedFrom SearchParameter http://hl7.org/fhir/SearchParameter/Organization-name, and this likely indicates that the derivation relationship is not valid
- The expression "Patient.id" is not compatible with the expression "Resource.id" in the derivedFrom SearchParameter http://hl7.org/fhir/SearchParameter/Resource-id, and this likely indicates that the derivation relationship is not valid
- The expression "Condition.onset.as(dateTime)" is not compatible with the expression "Condition.onset.as(dateTime) | Condition.onset.as(Period)" in the derivedFrom SearchParameter http://hl7.org/fhir/SearchParameter/Condition-onset-date, and this likely indicates that the derivation relationship is not valid
- The expression "Encounter.id" is not compatible with the expression "Resource.id" in the derivedFrom SearchParameter http://hl7.org/fhir/SearchParameter/Resource-id, and this likely indicates that the derivation relationship is not valid
- The expression "Location.name" is not compatible with the expression "Location.name | Location.alias" in the derivedFrom SearchParameter http://hl7.org/fhir/SearchParameter/Location-name, and this likely indicates that the derivation relationship is not valid
Eric Haas (Jul 02 2020 at 16:59):
thanks!
Eric Haas (Jul 03 2020 at 01:16):
I am unclear should I still be getting errors? because I am still getting these on my local build - deleted the FHIR cache and downloaded the latest build...
Grahame Grieve (Jul 03 2020 at 01:17):
sure you should still be getting those warnings - do you think they're wrong?
Eric Haas (Jul 03 2020 at 01:19):
to be clear I am seeing
SEARCHPARAMETER_EXP_WRONG
should I see something else?
Grahame Grieve (Jul 03 2020 at 01:19):
yes you should. What version is running?
Eric Haas (Jul 03 2020 at 01:28):
OK my downloader is not working properly
I'm still on Version 1.1.0 and I'm updating from here:
https://github.com/HL7/fhir-ig-publisher/releases/latest/download/publisher.jar
I thought that was the latest?
or is it here the link in the qa page:
https://storage.googleapis.com/ig-build/org.hl7.fhir.publisher.jar
Grahame Grieve (Jul 03 2020 at 01:30):
1.1.0 is the latest. @Mark Iantorno I do not understand why this error code is not converted - I think it must be a build/deploy problem - can you look into this? You need to run the IG publisher on the last US Core repo, which is https://github.com/HL7/US-Core-R4
Grahame Grieve (Jul 03 2020 at 01:31):
both of these locations have the same jar, but the second one will disappear next release (tmorrow)
Eric Haas (Jul 03 2020 at 01:31):
OK can you update the link in the QA page then too...
Grahame Grieve (Jul 03 2020 at 01:33):
yes thx
Mark Iantorno (Jul 03 2020 at 02:05):
I'll look into it first thing tomorrow morning. Are there dependencies within the code that read the version number directly?
Grahame Grieve (Jul 03 2020 at 02:38):
not sure what you're asking. But I think you mean can you rely on the fact that 1.1.0 is actually running? yes if the log says 1.1.10, then that's what's running for sure. See https://github.com/HL7/fhir-ig-publisher/blob/master/org.hl7.fhir.publisher.core/src/main/java/org/hl7/fhir/igtools/publisher/Publisher.java#L7646
Mark Iantorno (Jul 03 2020 at 15:22):
...how long is this process supposed to take?
Mark Iantorno (Jul 03 2020 at 15:22):
it's taking forever to run on my machine (>1 hour), is this normal?
Mark Iantorno (Jul 03 2020 at 17:12):
I'm getting our of memory exceptions trying to run this end to end
Mark Iantorno (Jul 03 2020 at 17:13):
I've increased the allocated memory for the jvm, but it's still not completing
Mark Iantorno (Jul 03 2020 at 17:15):
going to just give it a tonne and try again
Vassil Peytchev (Jul 03 2020 at 17:44):
It might be good to have a record of what "a tonne" of memory is in this case. From what I remember, 16 GB seemed to have been plenty before, so if it's more than that it's worth mentioning.
Mark Iantorno (Jul 03 2020 at 17:45):
I'm giving it 8GB now
Mark Iantorno (Jul 03 2020 at 17:45):
it's still running though
Mark Iantorno (Jul 03 2020 at 17:45):
How long does it take to run on average, against the us core ig
Rob Hausam (Jul 03 2020 at 17:52):
I don't know about US Core, but when I tried to build the IPS IG using Publisher 1.1.0 for the first time, it took 3 hours and 6 min.!
Mark Iantorno (Jul 03 2020 at 17:52):
Alright, so that makes sense for what I'm experiencing then
Mark Iantorno (Jul 03 2020 at 17:52):
thanks
Mark Iantorno (Jul 03 2020 at 17:52):
I wanted to make sure it wasn't my machine
Lloyd McKenzie (Jul 03 2020 at 17:53):
Is the txCache not committed?
Lloyd McKenzie (Jul 03 2020 at 17:53):
(It certainly should be...)
Mark Iantorno (Jul 03 2020 at 17:54):
So if I want to use the tx cache, how do I modify the basic command lloyd
Rob Hausam (Jul 03 2020 at 17:54):
it rebuilds the txcache - I still need to check the diff on it that Grahame mentioned to see what it's actually doing
David Pyke (Jul 03 2020 at 18:16):
Is it faster to have it rebuild the cache or wipe it and have it pull it all down again?
Rob Hausam (Jul 03 2020 at 18:37):
I expect that it will be about the same. But you could take several hours to test it both ways and report back the results (actually, I am curious about that). :)
David Pyke (Jul 03 2020 at 18:38):
Well, my system has a few spare cycles. Time to test it out and build USCore..
Rob Hausam (Jul 03 2020 at 18:40):
Sounds good. I will mention that we shouldn't be surprised (I won't be) if your mileage does vary. I expect that there will be some differences between IGs.
Eric Haas (Jul 03 2020 at 19:02):
Lloyd McKenzie: Is the txCache not committed?
it is indeed committed.
US Core should not take 3 hrs to build even the first time.
David Pyke (Jul 03 2020 at 19:03):
My first run (no txcache) Finished. 12:11.0582.
Rob Hausam (Jul 03 2020 at 19:04):
That's not bad at all (certainly in comparison to my > 3 hours).
David Pyke (Jul 03 2020 at 19:05):
Second run was Finished. 00:53.0574. I'm assuming I need to clear a build cache.
Rob Hausam (Jul 03 2020 at 19:05):
I think you need to checkout the previous set of cache files.
David Pyke (Jul 03 2020 at 19:06):
I'll try that
David Pyke (Jul 03 2020 at 19:07):
running with a fresh cloning
Rob Hausam (Jul 03 2020 at 19:07):
But it doesn't appear that you're likely to have much of a problem with US Core. I'm not sure why Mark's experience seemed so vastly different - but we'll see how this next run goes.
David Pyke (Jul 03 2020 at 19:08):
run with a fresh clone: Finished. 00:52.0571.
Rob Hausam (Jul 03 2020 at 19:09):
That's fast. There must be another variable.
Rob Hausam (Jul 03 2020 at 19:09):
The 3 hours was for IPS, @Eric Haas. But it shouldn't have taken maybe more than 10-20 minutes (tops), either.
David Pyke (Jul 03 2020 at 19:09):
Running it again, fresh download of the publisher, just in case
Rob Hausam (Jul 03 2020 at 19:12):
My second build of IPS with 1.1.0, following the txcache updates, took 5 minutes.
David Pyke (Jul 03 2020 at 19:14):
I think we need to chip in a buy Mark a new faster computer
David Pyke (Jul 03 2020 at 19:15):
I have noticed that validating some like ValueSet/us-core-narrative-status takes a significant time
Mark Iantorno (Jul 06 2020 at 13:06):
So I tried running it again over the weekend, while the computer wasn't doing anything else, and it still threw an OOM exception, despite allocating 16gb to the process... I'm going to talk with Grahame this week about it. I might be doing something incorrectly
Grahame Grieve (Jul 06 2020 at 19:20):
I don't know what you could be doing incorrectly, but it doesn't happen for me
Mark Iantorno (Jul 09 2020 at 13:05):
Just as an update to everyone here, this issue was fixed yesterday, and will be in the next release of the publisher.
Mark Iantorno (Jul 09 2020 at 13:10):
PR for the fix is here: https://github.com/HL7/fhir-ig-publisher/pull/127
Last updated: Apr 12 2022 at 19:14 UTC
