Dataset's runNumber inconsistencies

[minottic]

RunNumber in datasets is sometimes considered part of scientificMetadata others as a top dataset element. In the old backend it was part of scientificMetadata. The frontend treats it as top level element.

I would suggest we move it back to scientificMetadata (and the FE fetching accordingly), and have a migration which moves back

[bpedersen2]

This field was introduced by: #1485

Currently only the elasticsearch index uses the value from scientificMetadata

I see valid points for both places for that value to live. Maybe an configurable alias mechanism would work best, that allows to transparently map such fields onto each other?

[minottic]

thanks for digging out the PR!

TBH, I am also fine with any of the 2, and if it was reckoned better lived as top element, I am fine with it, I guess though it's yet another non backward compatible change, as the v3 response has changed compared to the old backend.

At least I believe a migration is missing? All our data annotated with runNumber in scientificmetadata is disregarded

[Junjiequan]

After the opensearch migration, the scientificMetadata field will no longer be indexed in the opensearch.

[minottic]

that's actually also used here. I believe we will still encounter inconsistencies though, especially from new users who see the DTO while old ones will use the scientificMetadata.runNumber. I am fine defining and running the migration from my side only, but I think a way to default as proposed in #2609 will prevent inconsistencies