Hi, I've opened and resolved https://issues.apache.org/jira/browse/FELIX-3017 in the trunk.
Regards, Clement On 30.06.2011, at 08:14, Clement Escoffier wrote: > Hi, > > On 28.06.2011, at 12:27, Thomas Calmant wrote: > >> Hi, >> >> I've a strange behavior while working on the iPOJO builder for Eclipse : >> when the iPOJO manipulator works on an already manipulated class, it returns >> an empty array. >> I've seen it comes from the method >> "org.apache.felix.ipojo.manipulation.Manipulator.manipulate(byte[])" >> (around line 70), where the test "ck.isalreadyManipulated()" is visible. >> >> Wouldn't be a better behavior to return the original byte array ("origin") >> instead of an empty one ? > > I think it's a former convention to determine is the class was manipulated, > but I agree it would be better to return the same array, and use the > isAlreadyManipulated method. > >> According to the Pojoization class, the setBytecode() method is called >> without tests, so it may erase the current class file content and imply an >> invalid build state... > > I'm not sure, but it should be tested... I will have a look to confirm it. > >> >> At least, it may be useful to indicate that behavior in the javadoc. > > Will do, > > Regards, > > Clement > >> >> Regards, >> Thomas. >