GCJ/Jacks test results for 02-05-2001

[Mo DeJong <mdejong at cygnus dot com>]

Hi all.

I just did a fresh checkout of the CVS sources
and re-ran gcj in the Jacks testing framework.
The last time I ran these checks (2000-12-03)
there were 1111 tests in Jacks, today
there are around 1600. Gcj currently passes
1276 of them. That compares with 1446
passes for Javac (from JDK 1.3),
1432 passes for the Jikes compiler,
and 1226 passes for the kopi compiler.

I did notice one possible regression.
The test case 14.20-catch-12 was
passing but now it seems to fail:

http://oss.software.ibm.com/developerworks/opensource/cvs/jikes/~checkout~/jacks/docs/tests.html#14.20-catch-12


This test looks like:

class Test {
        void foo() throws Exception {
               try {
                   throw new Exception();
               } catch (java.io.IOException io) {
                   // unreachable, as new cannot create subclass, and
                   // the constructor has no throws clause
               }
        }
}

The compiler should fail to compile this class, it
seems it was failing on 2000-12-03 but now it
incorrectly produces a .class file. Does that
ring a bell with anyone? I poked around
in the ChangeLog but nothing jumped out
at me as being directly related.

Mo DeJong
Red Hat Inc