Re: [aspectj-users] Aspects and overridden methods

[apel@xxxxxxxxxxxxxxxxxxxxxxx]

Hi Alex,

that makes sense to me. I am not that familar with AspectJ to know these
details. The consequence is that AspectJ is less robust against subsequent
included classes, etc. I have thought this is only a problem when adding
code that is inadvertently matched by an existing aspect that uses
wildcards, e.g.:

pointcut log() : call(* set(..));

Adding a subsequent class with a set method would be matched
unintentionally. But this can be partially avoided using naming
conventions (which is no guarrantee). But in the light of the above
example that joinpoints have multiple signatures this can cause additional
problems that cannot be so easily anticipated by a programmer.

sven

> Hi!
>
> According to JoinPoint Signature (described here [1])
>
> for method execution jp-s:
>
> [quote]
> There is one signature for each type that provides its own declaration of
> the method.
> [/quote]
>
> and considering the pointcut matching rules (stated here [2])
>
> [quote]
> # They are of the same kind
> # The signature pattern (exactly) matches at least one signature of the
> join point
> # The modifiers pattern matches the modifiers of the subject of the join
> point
> [/quote]
>
> I think that your execution pointcut matching is correct.
>
> Applying the same rules to the call pointcut, I think this is again
> correct.
>
> Do these make any sense to you?
>
> cheers,
>
> ./alex
> --
> .w( the_mindstorm )p.
>
>
>
>
> [1]
> http://www.eclipse.org/aspectj/doc/next/adk15notebook/join-point-signatures.html
> [2]
> http://www.eclipse.org/aspectj/doc/next/adk15notebook/jpsigs.html#join-point-matching
>
> #: apel@xxxxxxxxxxxxxxxxxxxxxxx changed the world a bit at a time by
> saying on  9/30/2005 7:38 PM :#
>> Hi,
>>
>> I have a problem with the following code example:
>>
>>  class Parent
>>  {
>> 	public void foo()
>> 	{
>> 		System.out.println("Parent::foo()");
>> 	}
>>  }
>>
>> class Child extends Parent
>> {
>>     public void bar()
>>     {
>> 		System.out.println("Child::bar()");
>>     }
>>
>> 	public void foo()
>> 	{
>> 		System.out.println("Child::foo()");
>> 		super.foo();
>> 		bar();
>> 	}
>> }
>>
>> aspect ParentAspect
>> {
>>     pointcut to_foo() : execution(void Parent.foo());
>>
>> 	after() : to_foo()
>> 	{
>>     	System.out.println("after calling: " +
>> thisJoinPoint.getSignature());
>> 	}
>> }
>>
>> When compling this example and executing the following main:
>>
>> public static void main(String[] args)
>> {
>> 	Child c = new Child();
>> 	c.foo();
>> }
>>
>> ... I get this output:
>>
>> Child::foo()
>> Parent::foo()
>> after calling: void Parent.foo()
>> Child::bar()
>> after calling: void Child.foo()
>>
>> Obviously the pointcut matches both foo() methods, although I wanted to
>> match only Parent.foo(). When trying a call instead the execution
>> pointcut
>> the result is that Child.foo() is matched only (which is clearly not
>> intended):
>>
>> Child::foo()
>> Parent::foo()
>> Child::bar()
>> after calling: void Child.foo()
>>
>>
>> For both cases (execution and called) I had expected that only the
>> Parent.foo() method was matched. Am I thinking wrong? To my opinion
>> there
>> must be a possibility to match only the Parent.foo() method without
>> using
>> the within pointcut, etc. (Imaging the Child class is added
>> subsequently,
>> in a later development step.)
>>
>> thanks in advance
>>
>> Sven
>>
>> _______________________________________________
>> aspectj-users mailing list
>> aspectj-users@xxxxxxxxxxx
>> https://dev.eclipse.org/mailman/listinfo/aspectj-users
>>
> _______________________________________________
> aspectj-users mailing list
> aspectj-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/aspectj-users
>