Oh, not a critique of startup or shutdown performance on T5-ioc, but
more of a general principle of unit testing components that
participate in any IoC container. To test the component, you
shouldn't need to use the container, because it's a "unit" test. But
he said he was testing the wiring between components anyway, rather
than the functionality of the units, so my comment isn't as relevant.
Christian.
On 4-Dec-08, at 15:23 , Howard Lewis Ship wrote:
> I use a mix of techniques, using a lot of mocks for true unit tests,
> but also a lot of integration tests.
>
> I'm not sure what CEG has actually seen here; Registry.shutdown() is
> very dramatic, it tears apart the registry (releasing almost
> everything to the GC) and informs all of the proxies to shutdown as
> well. Could he just be missing the re-creation of the services in
> later tests ... Registry startup is very, very fast once all the
> underlying classes are instantiated.
>
> On Thu, Dec 4, 2008 at 12:04 PM, Christian Edward Gruber
> <christianedwardgruber@gmail.com> wrote:
>> FYI, in general, you shouldn't be using the container in your
>> tests, unless
>> you're testing the wiring itself. You should be creating the
>> component/service under test, and constructing it with fakes. This
>> isn't
>> absolute but there is a lot more effort/configuration/overhead if
>> you want
>> to use the container infrastructure in your unit test, and you
>> start to have
>> subtle interactions that might potentially make it more of an
>> integration
>> test. You risk testing more than one thing at a time.
>>
>> Christian
>>
>> On 4-Dec-08, at 01:28 , Stephan Schwab wrote:
>>
>>>
>>> Hi!
>>>
>>> I have several JUnit tests that instantiate
>>> org.apache.tapestry5.ioc.Registry via the RegistryBuilder before
>>> tests
>>> run.
>>> Now I'm observing that services registered in one test are still
>>> available
>>> in other tests although I did call registry.shutdown(). My test
>>> runner
>>> does
>>> not fork a new JVM.
>>>
>>> Calling registry.shutdown() should cause everything to vanish. Is
>>> there
>>> anything that causes one-registry-per-JVM?
>>>
>>> Stephan
>>>
>>>
>>>
>>> -----
>>> --
>>> http://www.caimito.net - Caimito One Team - Agile Collaboration and
>>> Planning
>>> tool
>>> http://www.stephan-schwab.com - Personal blog
>>> http://code.google.com/p/tapestry-sesame - Authentication
>>> extension for
>>> Tapestry 5
>>>
>>> --
>>> View this message in context:
>>> http://www.nabble.com/IoC-registry-survives-between-JUnit-tests--tp20828078p20828078.html
>>> Sent from the Tapestry - User mailing list archive at Nabble.com.
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: users-unsubscribe@tapestry.apache.org
>>> For additional commands, e-mail: users-help@tapestry.apache.org
>>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@tapestry.apache.org
>> For additional commands, e-mail: users-help@tapestry.apache.org
>>
>>
>
>
>
> --
> Howard M. Lewis Ship
>
> Creator Apache Tapestry and Apache HiveMind
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tapestry.apache.org
> For additional commands, e-mail: users-help@tapestry.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tapestry.apache.org
For additional commands, e-mail: users-help@tapestry.apache.org
|