Schemas are broken

[jl-wynen]

The different dataset schemas are inconsistent with each other. As I understand it, you are in the process of changing them. But the current broken versions have been released and so affect Scitacean.

• The dataset class now uses lists for instrumentIds, proposalIds, sampleIds. But the creation DTOs still use the old scalar instrumentId, proposalId, samplesId, where the derived dataset only has proposalId.
• Investiagor:
  • Classs: principalInvestigator
  • Raw DTO: investigator and principalInvestigator
  • Derived DTO: investigator
• Type: An enum in the class and a plain string in the DTOs. Used to be an enum everywhere. Given that it is going to be more flexible, a plain string is probably best.

Can the above be fixed easily and quickly? Scitacean is currently broken and needs to be updated. But that requires a lot of manual hacks to account for the above inconsistencies. I would prefer to not do this since it would be temporary anyway.

[sbliven]

It looks like #1414 may have been merged prematurely and still hasn't been cleaned up. I'm a bit concerned that none of the tests revealed all the places where schema changes were not propagated. For instance, src/datasets/datasets.service.spec.ts still uses the old scalar properties, but the test passes and lints. VSCode flags some errors, but they don't seem to result in CI failures.

[nitrosx]

Being the author of PR #1414, I will check and fix it.

[nitrosx]

@jl-wynen the dataset class (aka the database representation of the dataset) correctly has only the principalInvestigator. Raw DTO should only have principalInvestigator. By the way, are you talking about create, update or output DTO?

To the SciCat user, the dataset type it is a string. Currently we only have raw and derived datasets. In the immediate future, site admins could add additional types to meet the needs of their institution.

[jl-wynen]

By the way, are you talking about create, update or output DTO?

Create. I did not check the others.

[nitrosx]

PR #1487 should address this issue

[jl-wynen]

Only partially.

• CreateRawDatasetObsoleteDto still has both principalInvestigator and investigator.
• The ids are the same scalars that they used to be in the schema. But the API returns the new array fields. So the generated models are wrong.

What is the prupose of OutputDatasetObsoleteDto? Is it even used anywhere?