I have a simple entity, Code, that I need to persist to a MySQL database.

public class Code implements Serializable {

    @Id
    private String key;
    private String description;

    ...getters and setters...
}

The user supplies a file full of key, description pairs which I read, convert to Code objects and then insert in a single transaction using em.merge(code). The file will generally have duplicate entries which I deal with by first adding them to a map keyed on the key field as I read them in.

A problem arises though when keys differ only by case (for example: XYZ and XyZ). My map will, of course, contain both entries but during the merge process MySQL sees the two keys as being the same and the call to merge fails with a MySQLIntegrityConstraintViolationException.

I could easily fix this by uppercasing the keys as I read them in but I'd like to understand exactly what is going wrong. The conclusion I have come to is that JPA considers XYZ and XyZ to be different keys but MySQL considers them to be the same. As such when JPA checks its list of known keys (or does whatever it does to determine whether it needs to perform an insert or update) it fails to find the previous insert and issuing another which then fails. Is this corrent? Is there anyway round this other than better filtering the client data?

I haven't defined .equals or .hashCode on the Code class so perhaps this is the problem.