d'oh -- now patch is attached On Fri, 30 Mar 2012, Yaroslav Halchenko wrote:
> patch is attached here as well -- =------------------------------------------------------------------= Keep in touch www.onerussian.com Yaroslav Halchenko www.ohloh.net/accounts/yarikoptic
From ebe8fbed067a95cfe03a8c900cfe9624201520fc Mon Sep 17 00:00:00 2001 From: Yaroslav Halchenko <[email protected]> Date: Fri, 30 Mar 2012 21:48:30 -0400 Subject: [PATCH] BUG: smirnov -- make comparison robust to handle NaNs any comparison (< or >=) with NaN is False, so check to belong to the range [x,y] must be done with (>=x)&&(<=y) and not !((<x)||(>y)) This issue happens to be critical especially for big-endian architectures leading to stalling in smirnov due to excessive looping later in the code. References: http://bugs.debian.org/cgi-bin/bugreport.cgi\?bug\=653948 http://mail.scipy.org/pipermail/scipy-dev/2012-March/017298.html --- scipy/special/cephes/kolmogorov.c | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-) diff --git a/scipy/special/cephes/kolmogorov.c b/scipy/special/cephes/kolmogorov.c index f587b6a..d78e0ae 100644 --- a/scipy/special/cephes/kolmogorov.c +++ b/scipy/special/cephes/kolmogorov.c @@ -35,7 +35,9 @@ smirnov (n, e) int v, nn; double evn, omevn, p, t, c, lgamnp1; - if (n <= 0 || e < 0.0 || e > 1.0) + /* This comparison should assure returning NaN whenever + e is NaN itself. In original || form it would proceed */ + if (!(n > 0 && e >= 0.0 && e <= 1.0)) return (NPY_NAN); if (e == 0.0) return 1.0; nn = (int) (floor ((double) n * (1.0 - e))); -- 1.7.8.3
_______________________________________________ Python-modules-team mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/python-modules-team
