RE: Method not found although apparently exists

["Eric Bodden" <eric@bodden.de>]

Hello!

First of all: Actually I attached a zip file containing all the class
files I was working on.

[Setting soot classpath explicitly...]
> Does that fix your problem?
No, unfortunately not at all since I am _always_ including the
sootclasspath explicitly anyway. And I _always_ include _myself_
%JAVAHOME%/bin/rt.jar.

> As for the feature of automatically adding rt.jar to the 
> soot-classpath if
> it is obtained from the $CLASSPATH environment variable, perhaps it
> should only do so if there is no rt.jar in $CLASSPATH, or perhaps it
> should be removed entirely. It is always possible to override it by
> specifying --soot-classpath explicitly. I welcome comments on this.
Yes, I think you should check if the rt.jar is already in there.
 
> I do not believe that this is a Solaris-only problem.
Me neither actually...
 
> > Second questions:
> > When such an error occurs, is it possible to let SOOT somehow ignore
> > this error and to tel it something like "OK, continue 
> anyway!" ? Or is
> > no CHA possible with such an error.
> 
> Phantom references were designed to do this. However, even with the
> --allow-phantom-refs switch, at that particular point, 
> phantom refs are
> forbidden. Could the person who wrote the code forbidding them there
> (allowsPhantomRefs field in Scene) please explain why? 
Wasn't me, but anyway: Even when I activate phantom classes (or
_deactivate_ this strange "alwaysDoNotAllowPhantomRefsWhatever"), I
always get the strangest exceptions when applying CHA. So am I right
that CHA on a Scene containing phantom classes will never work?

Thanks a lot,

Eric