Michael Paquier wrote, On 2015-03-25 01:19:
> Hence I'd rather think that patching src/port is the way to go, with
> for example something like the patch attached.
There are some corner cases that that patch does not handle properly,
most notably doubles that would overflow an integer. I suggest the
attached implementation (it's not as a patch, though). I've included a
test suite. The function can deal with non-IEEE floats too, but some of
the tests assume IEEE doubles.