I think it’s okay taking forward the Platform 2.3.0 full release. I and @tendomart have made a number of tests with the Reference application distribution built on top of the beta version. I have tested the rest endpoints using the default demo data and I think we are good to go. A couple of more commits on the platform are probably also on their way due to the on-going work on the Data filter module. Do we want to rather give more time to include the developments for a full release? Your thought and feedback on this is very helpful. Thank you.
Ok, let’s start from the beginning.
Sorry, I got confused with the fact that you are releasing platform, and UAT-refapp is of course to be used only by refapp releases, not platform. We do not have an UAT environment for platform.
If you are attempting to release platform 2.3.0-beta, why would you redeployed refapp 2.9.0 to UAT-refapp? That’s very confusing. So let’s just fix builds that should exist in the first place.
This is not a valid URL If you have questions about our environments, please refer to docs
Great @mksd. Thanks very much. I hope to have the full platform release by Tuesday(tomorrow) the 5th of November. Special thanks to every one that has contributed to this cause.
But is there any reason we (the community) have no single current UAT/QA test enviroment for the platform??. we seem to have alot of servers for the reff app and no single one for the platform .
The reason is pretty simple. It wasn’t needed so far.
There are no automated tests in platform that require an external qa-platform environment; so no reason to create it. The qa-refapp should be only used by automated builds in refapp, which was already there.
The UAT-platform wasn’t created because no one cared enough to create the environment.
If there’s an actual demand and I get some help, it’s not hard. But then I need guarantees that it’s going to be updated regularly after releases (which is currently not happening with the mdsbuilder, our only platform environment - which is not a build one).
i was going to mention about mdsBuilder but you know that is strictly restricted to very few people and release managers , we shouldnt even talk about that
I think the demand is there , every thing we do depends on the platform , new fixes are added every now and then . I think that is demand enough
which help is needed exactly ?? i dont think we can fail to offer the help if specified.
We already have a CI build that updates docker image openmrs-distro-platform in docker hub. It would need someone to add deployments to deploy tag ‘uat’ tag to dockerhub; exactly like it’s done to uat-refapp.
From there, a new docker service needs to be created in our docker-ansible repo, and that needs to have all the data needed on start up. UAT-platform will be recreated every couple of days.
Only the Infra team can then deploy this application to a server, including monitoring.
The last mile is updating wiki docs, creating a build to reset data in the server, change deployment to update the server, and so on.
Hello all… we’ve been testing against the 2.3.0-SNAPSHOT and have run into some interesting failing unit tests in one of our builds that will need some debugging. I’m guessing it has been caused by the change to the Person hibernate mapping.
This test, which was previously passing, fails when run against 2.3.0-SNAPSHOT:
In this case, the “voided” patient is now returned in the query.
Note that we void the patient in the first part of the test, so this may only be a problem that manifests itself during testing, but needs further exploring.
Looking deeper, the following test in core still passes…
… but when I put a breakpoint at 1756 in debugger I’m seeing the following strange configuration which I suspect is causing the problem when running a search:
But when I run this test in Core 2.2.x I see the following (which is what I would expect):
Okay, so I see that we added a bunch of fields to the Patient object in order to apply the mappings as annotations instead of in a separate mapping file, and these fields are identical to fields already on BaseOpenmrsData which I assume is what is causing the problem: