Re: [aspectj-dev] New build errors in sept 11 release of ajc

[Jim Hugunin <lists@xxxxxxxxxxx>]

Macneil Shonle wrote:
> > I think this is the tightening of a compiler loophole rather than a newly
> > introduced bug - but since this program has been working since 1.0.6 I'd
> > like Jim to verify my assessment!
> >        protected pointcut loadExecutions( Key key ):
> >                ContainerLoader.containerLoads( *, ContainerDescriptor )
> > && args(key);
> >
> > (please let me know if your system works as intended with this change).

Unfortunately, this is a new bug rather than a loophole tightening.  The 
problem was caused by moving name binding in pointcut declarations to 
happen before declare parents are evaluated.  Because of this, the 
compiler doesn't know that ContainerDescriptor isa Key when resolving 
the ContainerLoader.containerLoads reference.

The change in ordering was made to fix a bug reported in declare error 
and declare soft whose pcds where being evaluated before name binding 
had happened in the pointcut declarations.  Unfortunately, declare error 
and declare soft are concretized at the same time as declare parents 
(and all other declares ;-), so this move also led to the regression 
noted above.

FYI - AspectJ's static typing rules in pcd's work the opposite way that 
a standard method call does.  This is because information is flowing out 
of the call rather than into it.  The containerLoads pcd promises that 
it will provide an instanceof the ContainerDescriptor class.  Users of 
this pcd can treat this object as if it was a ContainerDescriptor or any 
SUPER type without any typesafety issues.

-Jim