👍 for the tests

I am also considering renaming:
NativeSQLiteConnectionManager -> EncryptedSQLiteConnectionManager
WebSQLiteConnectionManager -> NotEncryptedConnectionManager

This perhaps makes it even more readable and more platform-independent.

Feel free to re-request if I'm wrong

It feels like it's the correct place for sqliteFilePath, it's much cleaner now

Generally, what do you think about renaming clearSensitiveData() to indicate it also creates a fresh, clean DB? E.g. something like clearDataAndRecreateDB()?

Alternative is to move implementation to .cpp file, but not sure if this is worth it, I think the compiler will optimize it anyway, and I don't think this is worse in terms of readability.

This is temporary to make refactoring possible, next diffs could either not build or have incorrect logic without this. Later in the stack, this is updated to be private again.

This slightly affects ENG-10519 but makes it easier. I'd still decouple plain key generation from storing it in SecureStore, but it makes more sense for me to do it when working on rekeying

Thanks, this makes ENG 10521 trivial

I agree this was misleading, I was confused too when scoping rekeying the database.

Perhaps I should revert this in D14675

This only partially resolves issue with optional fields and their order, they still have to be processed after get_aux_data(), and get_aux_data() has to be called after any forward_field_to_blob() calls.
A proper solution requires redesigning multipart handling from scratch and is described in ENG-10773

Use camel case for version info

Fix attrs

Typo

Typo

Apply fix

Use alternative syntax

It's always satisfying to see this much of code being removed

Nice!