Stream: implementers
Topic: Montreal Freeze
Grahame Grieve (Mar 31 2016 at 11:39):
For the few who care (!), the pascal reference implementation has been updated to the frozen build at https://github.com/grahamegrieve/fhirserver/tree/master/reference-platform
Grahame Grieve (Mar 31 2016 at 11:39):
and http://fhir3.healthintersections.com.au is now running the current resource definitions, though I haven't yet updated it for the changes to the API
James Agnew (Mar 31 2016 at 12:26):
FYI to anyone interested, we'll have a Montreal build of HAPI FHIR ready by the end of the week too.
David Hay (Mar 31 2016 at 16:53):
@James Agnew - will you build a new CLI based on that version as well?
James Agnew (Mar 31 2016 at 17:37):
For sure.
David Hay (Mar 31 2016 at 18:44):
Brian Postlethwaite (Mar 31 2016 at 22:10):
The dotnet library will also be ready soon too.
Richard Ettema (Apr 06 2016 at 01:09):
All, the AEGIS FHIR server and client has been updated with support for "FHIR DSTU 2.2" - the Connectathon 12 snapshot. In addition, the previous two versions are also available for FHIR DSTU 2.0 and FHIR DSTU 2.1. The server base urls are:
AEGIS FHIR DSTU 2.0 Server --> http://wildfhir.aegis.net/dstu2 (no change)
AEGIS FHIR DSTU 2.1 Server --> http://wildfhir.aegis.net/dstu21 (updated url)
AEGIS FHIR DSTU 2.2 Server --> http://wildfhir.aegis.net/fhir (latest FHIR version for Connectathon 12)
The client urls are:
AEGIS FHIR DSTU 2.0 Client --> http://wildfhir.aegis.net/dstu2gui (no change)
AEGIS FHIR DSTU 2.1 Client --> http://wildfhir.aegis.net/dstu21gui (updated url)
AEGIS FHIR DSTU 2.2 Client --> http://wildfhir.aegis.net/fhirgui (latest FHIR version for Connectathon 12)
Pascal Pfiffner (Apr 07 2016 at 09:00):
FWIW, our Python and Swift clients now sport the 1.4.0.8139 data models. We don't plan to make this an official client release, those interested can use the develop branch on the respective repository.
In addition, the Python data models now perform a certain level of validation during deserialization and serialization.
https://github.com/smart-on-fhir/client-py/tree/develop
https://github.com/smart-on-fhir/Swift-FHIR/tree/develop
James Agnew (Apr 07 2016 at 11:34):
HAPI's DSTU3 endpoint has been launched too: http://fhirtest.uhn.ca/baseDstu3/
..or by testing UI: http://fhirtest.uhn.ca/home?serverId=home_21
Dave Carlson (Apr 07 2016 at 18:04):
Awesome, thanks Pascal!
David Hay (Apr 07 2016 at 18:23):
hate to say it, but http://fhirtest.uhn.ca/baseDstu3/metadata returns an error....
David Hay (Apr 07 2016 at 18:24):
here's the error returning the conformance through the UI: Error: org.hl7.fhir.dstu3.model.Conformance cannot be cast to ca.uhn.fhir.model.base.resource.BaseConformance
James Agnew (Apr 07 2016 at 22:37):
Thanks @David Hay , fixed. :)
And now of course I notice the the conformance statement for the DSTU3 endpoint has the wrong FHIR version coming out. Argh.
David Hay (Apr 08 2016 at 00:23):
good oh - let me know when you've deployed it...
Brian Postlethwaite (Apr 08 2016 at 04:58):
For those testing servers, mine is now available (not up to its usual standard, but should be ok)
http://sqlonfhir-may.azurewebsites.net/fhir
Vadim Peretokin (Apr 08 2016 at 07:42):
@Brian Postlethwaite it doesn't seem to be responsive
Jason Walonoski (Apr 08 2016 at 12:59):
@James Agnew HAPI DSTU3 seems to throw 500 errors if I specify a Content-Type on the request.
James Agnew (Apr 08 2016 at 13:44):
Thanks @Jason Walonoski , something is going on with the FHIR validator it looks like... What are you doing when this happens? (create/update/etc)
James Agnew (Apr 08 2016 at 13:47):
Ah got it.. I managed to reproduce it. Error fetching metadata of all things eh
James Agnew (Apr 08 2016 at 15:28):
Ok, that issue is solved. Thanks for letting me know!
David Hay (Apr 08 2016 at 23:23):
A reminder about the Montreal Connectathon - coming up soon! https://fhirblog.com/2016/04/08/montreal-connectathon/
Jason Walonoski (Apr 11 2016 at 14:42):
Crucible have beta STU3 models and client available for the Ruby programming language.
- https://github.com/fhir-crucible/fhir_client/tree/stu3
- https://github.com/fhir-crucible/fhir_models
Igor Sirkovich (Apr 11 2016 at 17:54):
Hello, we've been trying to migrate our Java server code from FHIR 1.0.2 to 1.4.0, but our developers have just reported some issues with the Java code they had downloaded from https://www.hl7.org/fhir/2016May/fhir-1.4.0-Java-1.0.zip. The firsts issue is that all the classes have a package name under "org.hl7.fhir.dstu3.formats" while the java classes are actually under "org.hl7.fhir.instance.formats" folder. Another issue, once they manually fixed the first one, is a missing "org.hl7.fhir.dstu3.exceptions" package.
Richard Ettema (Apr 11 2016 at 19:30):
@Igor, I track this issue down to the JavaGenerator class in the build/tools/java/org.hl7.fhir.tools.core project. The zip.addFiles method calls need to be updated with the correct path in the generated ZIP file. I have corrected and committed the necessary changes to the FHIR trunk. I'll let Grahame take care of the 2016May tag as the Java ZIP will need to be re-generated and uploaded to the STU3 site.
Igor Sirkovich (Apr 11 2016 at 19:48):
Thank you Richard! We'll keep working with our manually fixed version for now, but will switch to the FHIR one as soon as it gets reuploaded.
Richard Ettema (Apr 11 2016 at 20:26):
All, the AEGIS FHIR testing platform, Touchstone - http://touchstone.com, has recently been updated with support for the current FHIR STU3 Candidate version (v1.4.0-8139) which we have internally code named DSTU 2.2 which will be used at the upcoming FHIR Connectathon 12 event. We also have and will continue to retain support for the FHIR DSTU 2.0 (current official) and FHIR DSTU 2.1 (Connectathon 11) versions.
Richard Ettema (Apr 11 2016 at 20:26):
Connectathon 12 Test Definitions (TestScripts and corresponding fixture resources) for the following Test Tracks are now available:
- Track-06 DAF - ported and updated from Connectathon 11
- Track-08 FHIR Genomics - initial base operations tests for Observation and Sequence resources
- Track-13 Patient - ported and updated from Connectathon 11
- Track-16 Terminology - ported and updated from Connectathon 11
Richard Ettema (Apr 11 2016 at 20:27):
I will be assisting, as needed, the Test Track Leads and Coordinators with the preparation of TestScripts. I will also be uploading new/updated Test Track TestScripts as they become available to the Touchstone environment for pre-Connectathon testing and eventual use at the Connectathon 12 event.
James Agnew (Apr 11 2016 at 20:28):
I feel like the use of the terms "DSTU 2.1" and "DSTU 2.2" is potentially confusing to implementers, given that neither of these actually exist. Should that be updated to say "DSTU 3" for consistency?
Vadim Peretokin (Apr 11 2016 at 20:29):
I agree, it is confusing. A matrix might help here.
Richard Ettema (Apr 11 2016 at 20:40):
"DSTU 2.1" was the term used for the Connectathon 11 event so I'd like to keep that. We decided to use "DSTU 2.2" when we started our work on incorporating support for the Connectathon 12 event back in February. These snapshots, even though not official releases, are still versions in their own rights as people build their servers and clients support these testing events. That being said, I do agree it is a bit confusing. However, instead of "STU 3", this new version should be named "STU 3 Candidate" as it isn't the official "STU 3" release yet. Comments?
Vadim Peretokin (Apr 11 2016 at 20:41):
That would be nice and clear.
Richard Ettema (Apr 11 2016 at 20:46):
Ok. I'll work on updating our use of "DSTU 2.2" to "STU 3 Candidate". Thanks.
James Agnew (Apr 11 2016 at 20:51):
Why "STU" and not "DSTU"?
Vadim Peretokin (Apr 11 2016 at 20:52):
DSTU is no longer a thing per last WGM in January. Can't find a link right now
James Agnew (Apr 11 2016 at 20:52):
really? see.. i'm among the confused by all this.
Grahame Grieve (Apr 11 2016 at 20:53):
HL7 agreed to drop the "d" in dstu
Grahame Grieve (Apr 11 2016 at 20:53):
I've not implemented that in the java code yet
Hussain Chinoy (Apr 11 2016 at 20:53):
Congrats
Grahame Grieve (Apr 11 2016 at 20:53):
not sure whether I'm going to, or when
James Agnew (Apr 11 2016 at 20:53):
We've been running a matrix of FHIR versions to HAPI versions for the last few releases... There really isn't a nice way of presenting this info, it's confusing no matter how you slice it: http://jamesagnew.github.io/hapi-fhir/download.html
Grahame Grieve (Apr 11 2016 at 20:53):
STU 3 candidate is correct, I think
Vadim Peretokin (Apr 11 2016 at 20:55):
It should be easier to find a link to the announcement, but even the FHIR director didn't mention it in his blog *cough*
Igor Sirkovich (Apr 12 2016 at 03:06):
@Richard Ettema , I'm wondering if this fix will also resolve the issue with the missing "org.hl7.fhir.dstu3.exceptions" package. I'm also wondering if I should raise this as a tracker item.
Igor Sirkovich (Apr 12 2016 at 03:09):
Another issue our developers have discovered relates to JavaClass Generator for “org.hl7.fhir.utilities.ucum.special” package, e.g. Class ‘CelsiusHandler.java’ is under under org.hl7.fhir.utilities.ucum with class name specialCelsiusHandler.java’ rather than under org.hl7.fhir.utilities.ucum.special.
Grahame Grieve (Apr 12 2016 at 03:13):
Richard's fix fixes the first. I just fixed the second
Grahame Grieve (Apr 12 2016 at 03:13):
will post an updated version soon
Lloyd McKenzie (Apr 12 2016 at 03:28):
Technically both the Jan release and the May release are STU 3 candidates. Might want to differentiate and say "STU 3 May Candidate" - that'll work so long as we don't take too long getting STU 3 out the door . . . :)
Igor Sirkovich (Apr 12 2016 at 04:32):
Thank you @Grahame Grieve !
Brian Postlethwaite (Apr 12 2016 at 04:42):
Could also consider the naming to be specific to what it is for, the connectathon
e.g. STU3 c11 (which would apply for May)
Brian Postlethwaite (Apr 12 2016 at 04:42):
Then finish off with STU3 rtm ;)
Grahame Grieve (Apr 12 2016 at 20:13):
I have released a new version of the notepad++ plug-in which uses the latest version of FluentPath at http://www.healthintersections.com.au/FhirServer/npp-install-1.0.19.exe (still based on DSTU 2 resources; I haven't figured out how to support more than one version, and I haven't decided when/if I'll upgrade it to STU3)
Grahame Grieve (Apr 13 2016 at 00:19):
ok. so, several announcements regarding the montreal freeze / connectathon:
- I've completed my work doing QA on the fluent path expressions
- I've committed updated invariants and search path expressions to the 2016May branch
- the server at http://fhir3.healthintersections.com.au has been updated
- if you want to get the latest conformance definitions to support your development for montreal, you can either build the 2016May branch, or get them from http://fhir3.healthintersections.com.au/index
- my server at http://fhir3.healthintersections.com.au is now generating the indexes completely off the fluent path expressions defined in the spec. This is the culmination of a lot of work by many of us, and it's now completely possible for any other server to do the same
- if you want updated java code (see other thread about the java reference implementatoin) or updated pascal code, you can get them from my server rather than the formal frozen copy. HAPI and DotNet users etc will have to wait for the maintainers to update their releases accordingly (though they may not be affected by changes only to the expressions).
this means that the technical base for the reference implementations is different to the official version on which the connectathon is based. The *only* differences are in the expressions, and a couple of fixes I made to XPath statements as I went. The structures, value sets, profiles etc and the narrative is the same, and the expressions have only been changed to match the intent of the official version. Any questions, feel free to ask
Grahame Grieve (Apr 13 2016 at 00:20):
Significant contributers to the long process of getting to auto-generated indexes: Josh, Brian, Lloyd, Ewout, Bryn + there were probably others
Vadim Peretokin (Apr 13 2016 at 00:21):
Document /index not found for http://fhir3.healthintersections.com.au/index, did you mean another url?
Grahame Grieve (Apr 13 2016 at 00:40):
oops I meant http://fhir3.healthintersections.com.au/index.html
Igor Sirkovich (Apr 13 2016 at 14:24):
@Grahame Grieve , the downloads on your server don't work, error "Document /fhir-1.4.0-Java-1.0.zip not found" for Java and also for other files.
Grahame Grieve (Apr 13 2016 at 20:12):
yes, it turns out my server doesn't host the zip files :-(
Igor Sirkovich (Apr 13 2016 at 20:19):
Is there any other place where we can download the fixed Java zip, e.g. gforge or wiki or we should wait until the 2016May page gets updated?
Grahame Grieve (Apr 13 2016 at 20:28):
I'm just figuring that out
James Agnew (Apr 13 2016 at 21:20):
FWIW, I know you guys aren't using HAPI's JAR for REST but the Montreal data models (the RI ones) are in its 1.5 SNAPSHOT JARs and they are up on the Maven repos now.
There are a good few people I know of who are using HAPI/RI's model but not the REST framework at this point, i.e using JAX-RS or RestEasy.
Grahame Grieve (Apr 17 2016 at 22:56):
I just committed an update to the Montreal branch, mostly quality/implementation work:
- finish fluentpath implementation in java
- rewrite the validator to be faster and more accurate
- implement a structure map conversion for CDA --> FHIR
Grahame Grieve (Apr 17 2016 at 22:56):
all the same stuff is in the trunk, but the definitions of the resources are slightly different in the trunk
Brian Postlethwaite (Apr 18 2016 at 00:04):
Did you get the Date addition stuff in that fluentpath implementation too?
Grahame Grieve (Apr 18 2016 at 00:05):
no
Grahame Grieve (Apr 18 2016 at 00:06):
it's not specified, but also, we have no expressions that would use it
Grahame Grieve (Apr 18 2016 at 05:00):
one more thing: if anyone is interested, there's a java class in the reference implemnetation that converts between DSTU2 and montreal freeze versions of the a resource
Grahame Grieve (Apr 18 2016 at 05:01):
specifically, org.hl7.fhir.convertors.VersionConvertor
Igor Sirkovich (Apr 19 2016 at 14:20):
I'm wondering if for ConsentDirective, developers are supposed to use Contract Java model (there is no ConsentDirective model in FHIR Java RI)
John Moehrke (Apr 19 2016 at 14:54):
Yes, Privacy Consent Directive (PCD) is an IG defining a Profile on Contract.
Igor Sirkovich (Apr 19 2016 at 17:06):
Thank you John!
Last updated: Apr 12 2022 at 19:14 UTC
