regression in qmtest_libstdcpp

I've just noticed a regression here testing against libstdc++.so.6 
generated from gcc-4.0.1. The tests:

/demangle/abi_examples/01.cc 1 FAIL
/demangle/abi_examples/02.cc 1 FAIL

Andrew pointed me to the following PR, which looks to be the same issue:

http://www.opengroup.org:8000/lsb/publicpr/PRView?PR=0080

Reviewing the gcc-4.0.1 source, I see the upstream tests have changed, 
comparing them to the qmtest source (same change in 01.cc and 02.cc):

-  verify_demangle("f", "error code = -2: invalid mangled name");
+  verify_demangle("f", "float");

Making the above change in the qmtest source tree and re-running 
build_binary_testsuite, they now pass in that environment (they were 
failing in the same manner as the binary test suite)

Testing against our 10.2 release with libstdc++.so.6 built from gcc-3.4.3, 
the tests pass.

Since both libs are libstdc++.so.6 I can't easily do like I did for 
LSB-2.0 and provide a compatible library for LSB purposes.

I don't claim to know enough about the tests to say whether it's TSD or an 
upstream issue, but perhaps this will provide some more info to help 
decide.

-- 
Stew Benedict
Mandriva
Mandrakesoft becomes Mandriva