Java generic inference is unsound
Andrei Alexandrescu
SeeWebsiteForEmail at erdani.org
Tue Aug 25 08:56:08 PDT 2009
Depending how you look at this, it is either a success or a failure of
formal methods in designing programming languages.
http://twofoos.org/content/java-type-system-holes/
I didn't know about this. Loopholes in Java have come up now and then,
but I thought they'd fixed them all. An older one (2001) was this:
http://fpl.cs.depaul.edu/Problem.java
Igarishi, Pierce and Wadler had shown (OOPSLA 1999) that Java's core
type checking is correct, but have not included a host of generics in
the proof. Apparently that's still a problem.
Andrei
More information about the Digitalmars-d
mailing list