more investigation needed, also the exact use case to reproduce.
The behavior was noticed when debugging that the calculated nullability based on the entries in the CSV file (e.g. 0.8 i.e. Nullable) did not match what was generated in the corresponding EEA file (e.g. NonNull) and also what was shown in the javadoc hover, respectively.
Therefore it is suggested to handle this potential bug only after the database refactoring ("one source for javadoc+eea only") has been completed.
more investigation needed, also the exact use case to reproduce. The behavior was noticed when debugging that the calculated nullability based on the entries in the CSV file (e.g. 0.8 i.e. Nullable) did not match what was generated in the corresponding EEA file (e.g. NonNull) and also what was shown in the javadoc hover, respectively. Therefore it is suggested to handle this potential bug only after the database refactoring ("one source for javadoc+eea only") has been completed.