Re: [hibernate-dev] Static constants in org.hibernate.engine.internal.Versioning
On 06/23/2011 05:42 PM, Gail Badner wrote:
I just ran into an issue where EntityBinding.getOptimisticLockMode()
returns an ordinal from a org.hibernate.annotations.OptimisticLockType enum that is
inconsistent with the static constants in org.hibernate.engine.internal.Versioning.
Jira issue: http://opensource.atlassian.com/projects/hibernate/browse/HHH-6362
I'm not clear on what conventions there are for enums that are source-specific.
My preference would be to:
1) Extract the static constants for version type from
org.hibernate.engine.internal.Versioning into a new enum that is an SPI,
org.hibernate.engine.spi.OptimisticLockMode.
2) Change EntityBinding.getOptimisticLockMode() and
EntityBindingState.getOptimisticLockMode() to return
org.hibernate.engine.spi.OptimisticLockMode.
3) Change sources to be responsible for converting from a source-specific value to
org.hibernate.engine.spi.OptimisticLockMode.
Aside from the enum name, I think this is all goodness. These are
distinctly different from LockModes and we should avoid that confusion.
I tend to like to use the term "strategy" with pluggable interfaces,
but I think OptimisticLockingStrategy is definitely descriptive here.
Anyone else with better names?
In the same vein, I see there is
org.hibernate.metamodel.binding.CascadeType, which is not source-specific. In that class,
there's a TODO to integrate it with org.hibernate.engine.spi.CascadeStyle.
I also see that AttributeBindingState.getCascadeTypes() returns Set<CascadeType>.
My preference would be to remove org.hibernate.metamodel.binding.CascadeType and to
change AttributeBindingState.getCascadeTypes() and
AbstractAttributeBinding.getCascadeTypes() to:
org.hibernate.engine.spi.CascadeStyle getCascadeStyle();
I think this is fine. The comment is Hardy's though, so lets let him
respond to see if he had something else in mind.
--
Steve Ebersole <steve(a)hibernate.org>
http://hibernate.org