Re: [hibernate-dev] API removal proposal: org.hibernate.resource.jdbc.ResourceRegistry#cancelLastQuery
Hi Gail,
thanks for checking all that!
And sure I won't delete it out of the blue in a micro, I'd deprecate
it first, make a JIRA for removal later, etc etc but first I'd like to
hear if someone would prefer me to stay away and leave it there: while
I'm failing to see a good reason for this method to exist I guess
someone could know better.
Thanks
On Fri, 16 Aug 2019 at 21:24, Gail Badner <gbadner(a)redhat.com> wrote:
Hi Sanne,
I've confirmed that current versions of WildFly and EAP do not call that
ResourceRegistry#cancelLastQuery.
ResourceRegistry was introduced in 5.0. As of 5.0.16, cancelLastQuery() is not used by
Hibernate ORM.
At least as far back as 4.2, SessionImpl#cancelQuery called
JdbcCoordinator#cancelLastQuery(), which deals with JdbcCoordinatorImpl#lastQuery (not
ResourceRegistry#cancelLastQuery).
I believe that an application could only reach ResourceRegistry through an SPI ; for
example, via
* SharedSessionContractImplementor.getJdbcCoordinator().getResourceRegistry (deprecated
since 5.2);
or
*SharedSessionContractImplementor.getJdbcCoordinator().getLogicalConnection.getResourceRegistry().
My preference is to follow our normal process for removing an SPI.
In general, we try not to remove SPIs within a minor release (e.g., 7.x). We normally
deprecate the method in 7.x. The soonest we would remove it would be in 7.x+1.
Regards,
Gail
On Fri, Aug 16, 2019 at 10:17 AM Sanne Grinovero <sanne(a)hibernate.org> wrote:
>
> Hi all,
>
> is this method needed?
>
> - org.hibernate.resource.jdbc.ResourceRegistry#cancelLastQuery
>
> It seems undocumented, untested, and everything compiles just fine if
> I just delete it - however it seems marked as API.
>
> But also, I suppose this should be invoked by a different thread as
> the one actually running the query? Which would imply this
> implementation would need to be threadsafe, but it's not.
>
> My primary reason to explore its removal though is that this method is
> in the way of some optimisations I'd like to apply to this class, so
> it would be quite useful for me to be able to remove it - assuming
> that it's no longer of interest to anyone else.
>
> Thanks,
> Sanne
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev(a)lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hibernate-dev