[jboss-as7-dev] Modularity is the spawn of Lucifer and a stinking donkey

[Scott Marlow]

On 05/23/2012 11:09 AM, Steve Ebersole wrote:
> On Wed 23 May 2012 09:59:35 AM CDT, Scott Marlow wrote:
>> With this approach, Hibernate Search/OGM/Custom-Integrator will only
>> work with Hibernate ORM 4.1.4 or later. As long as the SPI doesn't get
>> broken in Hibernate 5.x/6.x/7.x/8.x, it might continue to work in
>> additional future Hibernate versions also (assuming we keep the design
>> in place that long).
>
> There are a number of suggestions/approaches in discussion. So when you
> say "this approach", I have no idea which approach you are referencing.
> Thats why I try to name them :)
>
>
>> The other option on the table, is having a very intimate AS7
>> dependency on
>> org.hibernate.service.internal.BootstrapServiceRegistryImpl class. If
>> someone changes
>> org.hibernate.service.internal.BootstrapServiceRegistryImpl in
>> Hibernate 5.x, they won't even know that they broke the integration.
>> Even if they don't change the class but do change the package name,
>> that will break Hibernate running on AS7 (with
>> Search/OGM/Custom-Integrator).
>
> Well you are suggesting a "very intimate dependency" as well with the
> SPI contract. I am talking about bytecode manipulation that hooks into
> instantiation of BootstrapServiceRegistryImpl. Not sure what changes can
> cause that to break other than moving packages which effects your SPI
> contract approach as well.

Just don't move the SPI class, as that will break Hibernate running on AS7.

>
>
> --
> steve at hibernate.org
> http://hibernate.org