[PRL] AspectJ "feature"

[Jeffrey Palm]

John Clements wrote:

> ...
> 
> Is this something that Gregor et. al. consider "the right thing?"
> 

Gregor I think wanted this to be an error; et al. (or just al.) made it 
the way it is.

I agree with you, but this is very similar to the 'fragile super type' 
(maybe that's not the name) problem, but just in disguise.  You write an 
aspect that's using the method names of a class as it's interface, but 
this interface isn't explicit.  Java and all the rest have the same 
problem with non-purely-virtual methods.  To me these are join points to 
which advice can be attached; the aspects are subclass instances; point 
cuts are the method signatures in subclasses; and the advice is the 
implementations of this method in those classes.  If the supertype 
changes the name of that method, the pointcut is subclasses is no longer 
valid.  That should be an error, too.

Jeff

> john
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> PRL mailing list
> PRL at lists.ccs.neu.edu
> https://lists.ccs.neu.edu/bin/listinfo/prl


-- 
Jeffrey Palm --> http://www.ccs.neu.edu/home/jpalm