On Wed, Jun 03, 2020 at 08:11:17PM -0400, Dennis Lee Bieber wrote: > On Tue, 2 Jun 2020 12:26:16 -0500, Albert Chin > <python-l...@mlists.thewrittenword.com> declaimed the following: > > >I've built Python 3.8.2 on AIX 5.2, 5.3, 6.1, and 7.1. I am seeing > >different results for the following Python program: > > $ python3 -c "eps = 2.0 ** -53.0; tiny = 2.0 ** -1022.0; \ > >print ((1.0 - eps) / tiny * 4.0)" > > > >I get the correct result, 1.7976931348623157e+308, on AIX 5.2, 5.3, > >and 6.1. But, on 7.1, I get "inf". > > > >Anyone know where can I look in the Python source code to investigate > >this? > > Have you considered that it might be something in an underlying C > library (especially for the double-precision exponentiation)?
On Wed, Jun 03, 2020 at 08:44:47PM -0700, Miki Tebeka wrote: > > Anyone know where can I look in the Python source code to investigate > > this? > > Probably around > https://github.com/python/cpython/blob/master/Objects/floatobject.c Thanks to both of you. I applied some updated OS patches and the problem went away. -- albert chin (ch...@thewrittenword.com) -- https://mail.python.org/mailman/listinfo/python-list