https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118038
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |missed-optimization,
| |rejects-valid
--- Comment #11 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
> In addition, bit_cast FP64<->I64 also works, for example, where I can also do
> numerical nonsense.
That is because gccs internal representation of real can represent all values
held for IEEE 64bit float (and can handle all values of 128bit IEEE float too).
It is just these semi-igregual formats it has issues with. Yes gcc can be fixed
but nobody is working on it right now.
If you want to fix it, real.{h,cc} are the files which are needed to start.
