You might know that Octave crashes on floating point exceptions
on a Dec Alpha. This is getting very annoying: overflows, underflows,
divisions and log of zero, etc, even some assignment operations
(a=1e-305) can crash Octave. Anybody have a solution? John, I know
that there is a problem with gcc, but would it be possible to either:
1) Trap FP errors and just return an error condition, or
2) Save the workspace before crashing?