Joe Buck wrote: > I've confirmed your bug report. It appears to be a regression, in > that your test case works correctly for gcc-3.4.2 but fails > for gcc 4.1.2. > Ok I'll fix it. Thanks, Paolo.