Make ValidationProblem implement Serializable

[iantabolt]

See #436 for more information.

Just a quick fix to make ValidationFailed exceptions actually serializable.

[havocp]

In discussion of #288 it looks like we are assuming ConfigOrigin is serializable (or at least, subclasses of it used in exceptions are). The transient here can't be the right fix in any case I don't think - we can't reconstruct this field after deserialization, and don't want to lose it.

[iantabolt]

Yeah I wasn't sure what to do with that, but looking at your commit here b739f4b, it seems like ConfigOrigin is very intentionally not serializable. Maybe I should follow the lead of ConfigException, and add custom writeObject/readObject methods that call ConfigImplUtil.writeOrigin and ConfigImplUtil.readOrigin?

[havocp]

ah, good archaeology, I didn't remember that at all.

I guess you are right - we should copy what ConfigException does with the custom read/write.

[iantabolt]

Ok I've updated it to use the custom serialization. I tried to extract a method to reuse some of the other code but maybe it would've been more straightforward to copy/paste the whole thing ¯_(ツ)_/¯. Let me know

[havocp]

Thanks for working on this!

[havocp]

Seems good! Mind adding a unit test or two as appropriate? There's hopefully an existing example for serializing exceptions and test for ValidationProblem could go next to it following a similar pattern. Or test an exception with problems in it.

[iantabolt]

Sure thing

[2m]

Looking good. Great that you have extracted the logic from the way ConfigException was handled. After adding a missing unit test, this can be merged.

[iantabolt]

Added a test. Sorry it took almost a year! 😬

[iantabolt]

Is this able to be merged?

[havocp]

Thanks! sorry for the delay...