Raymond Hettinger <raymond.hettin...@gmail.com> added the comment:
I had not put any thought into subnormals (denormals?) because they didn't arise in my use cases. But it would be nice to handle them as well as possible. Accuracy improvements are welcome as well. And that goes hand in hand with better commutativity. The frexp() approach looks cleaner than my big/little matching algorithm and it doesn't require auxiliary memory. It may slower though. The matching algorithm only kicks in when overflows or underflows occur; otherwise, it runs close to full speed for the common case. ---------- _______________________________________ Python tracker <rep...@bugs.python.org> <https://bugs.python.org/issue41458> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
