https://gcc.gnu.org/bugzilla/show_bug.cgi?id=48852
--- Comment #14 from Jerry DeLisle <jvdelisle at gcc dot gnu.org> --- I have decided that to do this right I need to revamp how write_float.def and associated functions are organized. We have discussed doing this quite a while ago, so now is the time. I see it like this: - Create a working buffer (currently buried down under, need it brought up) - Get the digits based on the KIND etc (DTOA FTOA etc) - Format the digits (right now mostly in output_float) - Send the formatted digits out (split this out of output float)