Force repo config #122

poikilotherm · 2023-01-21T16:33:44Z

@landreev said we better force people to configure everything, so hard to track bugs like IQSS/dataverse#9309 don't happen as easily.

- Make sure users of the library do add critical configuration like earliest date - We still have some sensible defaults e.g. for list sizes - Make it a requirement to set a configuration when creating the repository - As the configuration is sealed and non-modifiable now, retrieve a builder from it again to modify and inject into a repository instance - Add more tests for the configuration builder - Move the "defaults()" thing to the test class, as people should not be using it when implementing a repository

… in config builder

…nd its builder

…est date

landreev · 2023-01-23T15:04:11Z

Thank you! Will add review shortly. Since the PR is ready, we don't really need a separate issue requesting this change, do we? (And same for the other PR for the "earlier than the earliest datestamp"?)

poikilotherm · 2023-01-23T18:03:50Z

Ok 🤫 Don't tell anyone we did not create an issue first.

landreev · 2023-01-24T15:31:17Z

...ata-provider/src/main/java/io/gdcc/xoai/dataprovider/repository/RepositoryConfiguration.java

+        DeletedRecord deleteMethod;
+        Integer maxListIdentifiers = 100;
+        Integer maxListSets = 100;
+        Integer maxListRecords = 100;


Probably a stupid question, but why do we need these default values hard-coded here? - considering that we have microprofile config defaults for these?

This has been a hardcoded default before and is also recommended within the OAI PMH implementation guidelines, too.

Yes, Dataverse has its own defaults. But anyone looking to reuse this library might have a different opinion, etc.

I do not want to add another dependency on MPConfig to this library, so I think leaving the 100 in place is a reasonable compromise between defaults, backward compatibility and reusability.

In my defense, I did acknowledge early on it was a stupid question. (I was literally confusing what we do in the Dataverse code base and what you do in this library - sorry).