[infinispan-dev] Change AbstractInfinispanTest.nullifyInstanceFields method name to zzzNullifyInstanceFields

[Galder Zamarreno]

On 10/19/2009 03:46 PM, Vladimir Blagojevic wrote:
> You gotta be kidding! Are you sure that your method is annotated with
> @AfterClass and not @AfterTest. Is this in svn? I'd like to try it myself!

I'm not kidding and I'm sure :)

I've just committed it, r969. Do an svn updated, revert to the 
AbstractInfinispanTest.nullifyInstanceFields and run 
org.infinispan.jndi.BindingTest and you'll see an NPE being thrown in 
the preDestroy() method.

>
> On 09-10-19 7:55 AM, Galder Zamarreno wrote:
>>
>> On 10/19/2009 01:52 PM, Galder Zamarreno wrote:
>>
>>> Hi,
>>>
>>> I'm trying to modify SingleCacheManagerTest so that in the same way that
>>> there's a preCreate() callback, there can be a preDestroy() callback
>>> too. This is so that for example, in the
>>> https://jira.jboss.org/jira/browse/ISPN-153 test, I can stop the naming
>>> server created at in preCreate().
>>>
>>> The problem though is that by the time either
>>> SingleCacheManagerTest.destroyAfterClass is called,
>>> AbstractInfinispanTest.nullifyInstanceFields has already run and I get NPEs.
>>>
>> Sorry, here meant: before  SingleCacheManagerTest.preDestroy is called,
>> AbstractInfinispanTest.nullifyInstanceFields  has already run.
>>
>>
>>> However, I see other cases such as
>>> BdbjeCacheStoreFunctionalIntegrationTest where clearTempDir is called
>>> before AbstractInfinispanTest.nullifyInstanceFields.
>>>
>>> I just run a test without the preDestroy() in SingleCacheManagerTest so
>>> that there was only nullifyInstanceFields and destroyAfterClass
>>> @AfterClass methods to called and destroyAfterClass was called first.
>>> So, there does appear to be some kind alphabetical order in which testng
>>> calls its @AfterClass methods.
>>>
>>> So, I've just tried to change
>>> AbstractInfinispanTest.nullifyInstanceFields method name to
>>> AbstractInfinispanTest.zzzNullifyInstanceFields and now the preDestroy()
>>> gets called first.
>>>
>>> Any objections to this change?
>>>
>>> The other solution would to have some kind of dependsOnMethods in
>>> AbstractInfinispanTest but that would force specific naming on
>>> subclasses. Changing the name in AbstractInfinispanTest seems like a
>>> much easier option.
>>>
>>
>
> _______________________________________________
> infinispan-dev mailing list
> infinispan-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/infinispan-dev

-- 
Galder Zamarreño
Sr. Software Engineer
Infinispan, JBoss Cache