Classification on nested branches

[liquid36]

Hi! I am creating a new Snomed Extension and i want to use the classifier. My branch set up is:

• International Edition
• Spanish Extension
• Argentinian Extension

My branch path is: MAIN/SNOMEDCT-ES/SNOMEDCT-AR.

I created a new codesystem called TEST and set the metadata to the branch MAIN/SNOMEDCT-ES/SNOMEDCT-AR/TEST:

{
  "assertionGroupNames": "common-authoring",
  "defaultNamespace": "1000309",
  "internal": {
    "classified": "false"
  },
  "dependencyRelease": "20231130",
  "dependencyPackage": "SnomedCT_Argentina-ExtensionRelease_PRODUCTION_20231130T120000Z.zip",
  "defaultModuleId": "733690761000309105"
}

After created a new concept, i run the classification service via Snowstorm but it did not work as expected because the Argentina Extension zip file does not contain all the information.

Which should the proper configuration be for this set up? I tried to set the dependencyPackage to the International Edition Zip File and it worked fine. But, of course, it doesn't have the Argentinian concepts.

[kaicode]

Not all of the SNOMED International tooling is able to cope with an extension of an extension. It works well when maintaining an extension of an edition (just two packages). Are you maintaining the Argentinian Extension or creating an extension based on that? The simplest workaround is to create an Edition package containing all of the RF2 files that you are extending.

[liquid36]

Are you maintaining the Argentinian Extension or creating an extension based on that?

I am creating an extension based on Argentina Extension.

The simplest workaround is to create an Edition package containing all of the RF2 files that you are extending.

Yes, i think that's also.

thanks

[kaicode]

To create an Edition package that works with SI tooling it should be okay to just unzip all the RF2 txt files into one Snapshot folder and then create a zip from that. This would result in multiple Concept files but this should be okay. It should not be necessary to combine the rows of each file type. Please let me know if this approach works.

A good test is to classify using an empty delta archive with the edition package as the dependency. There should be no rows in the results RF2 files.

I use this "empty delta" for tests like this - empty-delta.zip.

[liquid36]

Okey, it's make sense to me. For now, I'm going to use the Argentina Edition Package.