Add support for ObjectId

[stIncMale]

HIBERNATE-23

[stIncMale]

This way it is easier to express what exactly we are expecting than with a string.

[NathanQingyangXu]

Looks great overall. Only some minor comments are left.

[NathanQingyangXu]

Just found an edge case wherein explicit @JavaType annotation is used, the getById() testing case would fail:

@Entity
    @Table(name = "items")
    static class Item {
        @Id
        ObjectId id;

        @JavaType(ObjectIdJavaType.class)
        ObjectId v;
    }

[stIncMale]

Regarding #70 (comment)

Just found an edge case wherein explicit @JavaType annotation is used, the getById() testing case would fail

This is because Hibernate ORM instantiates ObjectIdJavaType on its own in that case. Consequently, when it calls ObjectIdJdbcType.getBinder, the instance of ObjectIdJavaType passed as the javaType argument is not the same as ObjectIdJavaType.INSTANCE. I added the ObjectIdJavaType.equals/hashCode overrides to take this into account. However, applications can't explicitly use ObjectIdJavaType, ObjectIdJdbcType, because they are both internal. Are you aware of any reason to make them part of the API now?

Done in 60c7345.

[NathanQingyangXu]

Regarding #70 (comment)

Just found an edge case wherein explicit @JavaType annotation is used, the getById() testing case would fail

This is because Hibernate ORM instantiates ObjectIdJavaType on its own in that case. Consequently, when it calls ObjectIdJdbcType.getBinder, the instance of ObjectIdJavaType passed as the javaType argument is not the same as ObjectIdJavaType.INSTANCE. I added the ObjectIdJavaType.equals/hashCode overrides to take this into account. However, applications can't explicitly use ObjectIdJavaType, ObjectIdJdbcType, because they are both internal. Are you aware of any reason to make them part of the API now?

Done in 60c7345.

If they are internal, I think the edge case doesn't make sense, right? I missed that point. Should we revert back this commit then if it is not a concern any more?

[NathanQingyangXu]

Let us revert back the edge-case related stuff if these type classes are internal ones, not meant to be used externally

[NathanQingyangXu]

Let us revert back the edge-case related stuff if these type classes are internal ones, not meant to be used externally

It doesn't hurt to leave them as they are. It might improve our product robustness.

[stIncMale]

Let us revert back the edge-case related stuff

Talked about this, and agreed to leave the changes: despite applications not being able to use @JavaType(ObjectIdJavaType.class), this example gave us knowledge that Hibernate ORM may instantiate an implementation of JavaType on its own, even if it is not designed to be externally instantiated (for example, has a private constructor; note how Hibernate ORM does not do the same for entities, so this behavior is weird to me). Taking this behavior into account in our code seems better, even if we don't expect it. The alternative is to throw FeatureNotSupportedException, and leave an comment explaining what is going on. Having a test that gives an example seems better than having a comment.