Subject: [Fwd: Generic Java type inference is unsound]
From: Craig Chambers (chambers@cs.washington.edu)
Date: Mon Dec 17 2001 - 13:48:47 PST
Zowie!
-- Craig
Alan Jeffrey wrote:
>
> [----- The Types Forum, http://www.cis.upenn.edu/~bcpierce/types -----]
>
> Dear types researchers,
>
> Bracha, Odersky, Stoutamire and Wadler's Generic Java (OOPSLA 1998) is
> an extension of Java to include support for generic types and methods.
> It forms the basis of the Java Specification Request draft specification
> `JSR14: Add Generic Types to the Java Programming Language' described
> at:
>
> http://www.jcp.org/jsr/detail/14.jsp
>
> The core of the type checking system was shown to be safe by Igarishi,
> Pierce and Wadler (OOPSLA 1999) but the type inference system for
> generic method calls was not subjected to formal proof. In fact, it is
> unsound, as demonstrated by the program available at:
>
> http://fpl.cs.depaul.edu/Problem.java
>
> This program compiles without error, but produces a ClassCastException
> at run time, even though it contains no explicit casts. (The casts are
> produced by the compilation down to JVM bytecode, and should never
> fail.)
>
> This problem has been verified by the JSR14 committee, who are working
> on a revised langauge specification.
>
> Cheers,
>
> Alan.
>
> --
> Alan Jeffrey http://fpl.cs.depaul.edu/ajeffrey/
> CTI, DePaul University, 243 S. Wabash Ave, Chicago 60604, USA
_______________________________________________
Cecil mailing list
Cecil@cs.washington.edu
http://majordomo.cs.washington.edu/mailman/listinfo/cecil
This archive was generated by hypermail 2b25 : Mon Dec 17 2001 - 13:49:04 PST