Test Cases For Custom Date Serialization

[seamuslowry]

I am new to the open source world and this is the first time I'm writing any groovy. Please attribute any mistakes to ignorance rather than malice

[bartoszwalacik]

Good test case. Custom JSON serializers should have priority over built-in serializers. Since you have already found the place in javers-core where serializers are beeing registered, think about contributing a PR with the fix. But remember, custom serializers manage only how your data are stored in JaversRepository. Diff logic is customized in CustomComparators.

[seamuslowry]

It may take a couple days as I'm a little pressed for time right now, but I think I could do that. Would it be appropriate to just keep committing to this pull request?

[bartoszwalacik]

take your time,
yes you can commit here

[seamuslowry]

What do you think of that as a solution? It would be tedious to write the test cases such that they test all the adapters in Java8TypeAdapters and UtilTypeCoreAdapters for the ability to override, but I'll do it if you feel it is necessary.

I tried to generalize it but a lot of the types don't have a constructor valid for something like clazz.getDeclaredConstructor().newInstance() so I couldn't do a full generalization.

As it is, at least one type from each of Java8TypeAdapters and UtilTypeCoreAdapters is tested to ensure it can be overridden and I threw in a test to override the adapter for Atomic as well.

[seamuslowry]

It wasn't sitting well with me to leave cases untested, so I wrote tests to verify that the 17 built in adapters I identified and affected could be properly overridden. I tried hard to generalize the tests so there wouldn't need to be 17 new adapter classes for this test, but I couldn't get to a solution that both tested all 17 cases and was fully generalized.

[bartoszwalacik]

I'm on holiday, will back to you after a few days

[bartoszwalacik]

Indeed, these 17 new adapter classes in the test case are not nice

[bartoszwalacik]

I have fixed this test using composition, see https://github.com/javers/javers/pull/942/files

[seamuslowry]

Clever. I never thought of using the real adapters as delegates. I'll keep that pattern in my back pocket. Thanks!

[bartoszwalacik]

👍 thanks for your contribution
this fix just released in 5.8.10 https://javers.org/release-notes