Mark Dickinson added the comment: I'm thinking of something like the attached patch, which limits nd to 2 billion, comfortably within the range of int on machines that Python's likely to meet. (If anyone's worried about machines with ints smaller than 32 bits, we could add a configure check for that.)
I don't think we can take the limit all the way to 2**31 - 1, since nd may be combined with the exponent (which is limited to less than 20000 in the current code), but a limit of 2 billion should be safe. With this limit in place, it should then be safe to silence the warnings. ---------- Added file: http://bugs.python.org/file32810/limit_dtoa_string.patch _______________________________________ Python tracker <rep...@bugs.python.org> <http://bugs.python.org/issue19638> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
