Python doesn't know how to do math?
John Baxter
jwbaxter at spamcop.net
Fri Apr 26 11:05:51 EDT 2002
In article <gb4hcu4b55gq99blg08eudh14pm8ivju0k at 4ax.com>,
Chris Spencer <clspence at one.net> wrote:
> I want to thank everyone who responded to this. I read the floating
> point tutorial and now understand the issues.
And now...just a wee reminder that it is possible for the chip makers to
build errors into their floating point processors. Even large,
well-known chip makers whose names start with "Intel."
And it's possible for software writers to write bugs into their floating
point software routines. Even large companies deeply involved in the
IEEE standards process and producing an excellent implementation of that
standard...including one whose name starts with "Apple" (whose code in
rare cases messed up the low bit of the mantissa).
So...while the issue you raised isn't the result of an error, that
doesn't mean that no strange result is ever an error. Keep questioning
oddities! Understanding flows therefrom.
--John
More information about the Python-list
mailing list