Re: [gfortran,patch] Add option to raise FPE

[François-Xavier Coudert <fxcoudert at gmail dot com>]

>> Why one bitmask and not 6 arguments? The code seemed pretty readable
>> in that way. Plus, it enables us to have more than two states for each
>> exception, which might be useful at some point.
>
> Huh?  Like what?  It's not like the hardware has more than "on"
> and "off".  AFAICS, this only makes the code larger.

OK, I can reduce it. I thought about having the choice between
"sending a SIGFPE" or "doing nice things by catching this SIGFPE to
display a nice message", but it's probably too much.

>> Could you point me to one such occurence? A grep of the GCC codebase
>> did not help me much...
>
> config/gcc/crtfastmath.c.

Thanks! I'll look at that and submit a new proper patch.

As for the glibc feenableexcept, I doesn't support raising a FPE on
denormal numbers. I don't know whether we do really want that: I don't
think so, but I would be glad to have opinions from all people.

FX