[hibernate-dev] Dropping "Serializable" requirement for IDs ?

[Steve Ebersole]

Actually, I correct myself.  The spec does say that composite keys must be
Serializable (section 2.4).

I'd say that we cover this under our "portability/compliance guidelines"
like we do for other JPA requirements that we "softly" require.

On Wed, Jul 6, 2016 at 5:23 PM Steve Ebersole <steve at hibernate.org> wrote:

> I believe the spec actually just says that the composite key class must be
> serializable if the model is to be serialized (duh).  I could be
> misremembering, but I believe that is right.
>
> The spec does not say anything about the serializability of simple keys
> because the spec severely limits id types (which are inherently
> serializable).
>
> For the most part, this is our requirement.
>
> On Wed, Jul 6, 2016 at 4:10 PM Gunnar Morling <gunnar at hibernate.org>
> wrote:
>
>> Serializability of composite key types is mandated by the JPA spec, but I
>> suppose it's fine to be less strict in our implementation. Out of
>> interest,
>> does anyone know why that requirement is part of the spec?
>>
>> 2016-07-06 22:19 GMT+02:00 Sanne Grinovero <sanne at hibernate.org>:
>>
>> > Tracked now as:
>> >  - https://hibernate.atlassian.net/browse/HHH-10935
>> >
>> > Thank you all!
>> >
>> > On 6 July 2016 at 17:49, Emmanuel Bernard <emmanuel at hibernate.org>
>> wrote:
>> > > +1, one less people will complain about :)
>> > >
>> > > On Tue 2016-07-05 17:28, Steve Ebersole wrote:
>> > >> I'm ok with this.  That is no longer a real requirement.
>> > >>
>> > >> Any disagree?
>> > >>
>> > >> On Tue, Jul 5, 2016, 11:58 AM Sanne Grinovero <sanne at hibernate.org>
>> > wrote:
>> > >>
>> > >> > Hi all,
>> > >> > today creating a unit test I was greeted by this "old friend":
>> > >> >
>> > >> > > org.hibernate.MappingException: Composite-id class must implement
>> > >> > Serializable:
>> > >> >
>> > >> > shall we get rid of this requirement?
>> > >> >
>> > >> > It's just an inconvenience for end users to have to remember this
>> > >> > marker, as far as I know we don't really need this anymore.
>> > >> >
>> > >> > I realise some code might still be needing this, but we probably
>> can
>> > >> > do better by not actually serializing it with the default platform
>> > >> > serializer?
>> > >> >
>> > >> > Just an idea for a potential 6.0 improvement..
>> > >> >
>> > >> > Thanks,
>> > >> > Sanne
>> > >> > _______________________________________________
>> > >> > hibernate-dev mailing list
>> > >> > hibernate-dev at lists.jboss.org
>> > >> > https://lists.jboss.org/mailman/listinfo/hibernate-dev
>> > >> >
>> > >> _______________________________________________
>> > >> hibernate-dev mailing list
>> > >> hibernate-dev at lists.jboss.org
>> > >> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>> > _______________________________________________
>> > hibernate-dev mailing list
>> > hibernate-dev at lists.jboss.org
>> > https://lists.jboss.org/mailman/listinfo/hibernate-dev
>> >
>> _______________________________________________
>> hibernate-dev mailing list
>> hibernate-dev at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>>
>