Clearly something is not right if you get different results in AJDT and outside AJDT for the same codebase. Can you turn pipeline compilation off in both cases and see if that at least makes behaviour consistent? I think you should raise a bug for the inconsistency and we'll look at it.
I have fixed bugs vaguely in this area but that was a few months ago so the fixes should be in both the AJDT and AspectJ you are using.
I tried recreating your problem in a small example and it behaves fine for me, so if you are able to distill it to a small program that fails and attach that to the testcase, that will help immensely.
cheers,
Andy.
2008/8/18 Guillaume Pothier
<gpothier@xxxxxxxxx>
Hi,
I used Eclipse 3.4/AJDT (1.6.0.200808111932) to develop my AspectJ
program. It compiles fine using the integrated compiler. However when
I try to build the same program with the iajc ant task (from AspectJ
1.6.1), some of my declared errors get triggered.
More specifically, I have an annotation that I want to be able to
apply only on methods of classes that implement a particular
interface. So I have this declaration in an aspect:
declare error:
execution(@Scheduled * (! IJobSchedulerProvider+).*(..))
: "@Scheduled method must belong to a IJobSchedulerProvider";
There is no violation of this constraint in my program, and this works
well in AJDT.
However, iajc reports an error if a method has the annotation and
belongs to an abstract class, even if the class does implement the
interface. Is that a bug, or am I doing something that has borderline
semantics?
Regards,
g
_______________________________________________
aspectj-users mailing list
aspectj-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/aspectj-users
