Hi,
Given the discussion yesterday, I played around with some changes on a
branch when I got up early this morning to show how we could do a
partial migration to Spring without having to do it all at once.
https://svn.apache.org/repos/asf/maven/archiva/branches/springy
This shows:
- ability to lookup plexus components via spring IoC
- ability to lookup spring beans during the Plexus component lifecycle
- basic functional setup for Spring in the Archiva application
Eventually, as whole subsystems no longer require plexus it will be
possible to clean it up, such as:
- get rid of the additional lookups
- use annotations for configuration
- use testng + get/set + mocks for the tests where possible (and
spring testcontext where integration testing is needed)
Here is how to obtain a plexus object from Spring (note there is some
pre-req setup in test cases you'll see in the commit, as there is in
the additional servlet listener):
<bean id="urlCache" factory-bean="plexusCacheFactory" factory-
method="createInstance" />
<bean id="plexusCacheFactory"
class="org.apache.maven.archiva.common.spring.PlexusFactory">
<constructor-arg index="0"
value="org.codehaus.plexus.cache.Cache"/>
<constructor-arg index="1" value="url-failures-cache"/>
</bean>
To get a spring bean inside a plexus component, it is like this (make
sure to implement Initializable):
/**
* @plexus.requirement
*/
private SpringFactory springFactory;
public void initialize()
throws InitializationException
{
urlFailureCache = (UrlFailureCache)
springFactory.lookup( "urlFailureCache" );
}
The next thing we should probably try is using something like
SpringCache as suggested to remove the plexus-cache dependency.
Have fun!
Cheers,
Brett
--
Brett Porter
brett@apache.org
http://blogs.exist.com/bporter/
|