> I know it was my fault and that it would have been my job to
> fix it. But since I've installed glibc-2 (libc6) and
> gcc-2.8.1, many regressions fail due to floating point and
> error message diff's.
Yeah, that seems to be a problem (or at least an annoyance). I've got
egcs-2.91.57 installed, but for running regression tests I go back to
gcc-2.7.2.1 to get the rounding behavior back as it used to be.
- Tom