[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Octave 3.6.1 Make Check on FreeBSD
From: |
marco atzeri |
Subject: |
Re: Octave 3.6.1 Make Check on FreeBSD |
Date: |
Wed, 29 Feb 2012 00:23:50 +0100 |
User-agent: |
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:10.0.2) Gecko/20120216 Thunderbird/10.0.2 |
On 2/28/2012 9:19 PM, Thomas D. Dean wrote:
scripts/plot/axis.m ............................ PASS 1/2 FAIL 1
scripts/statistics/distributions/logninv.m ..... PASS 15/19 FAIL 4
scripts/statistics/distributions/nbincdf.m ..... PASS 14/20 FAIL 6
scripts/time/datevec.m ......................... PASS 11/12 FAIL 1
test_system.m .................................. PASS 95/96 FAIL 1
Summary:
PASS 9975
FAIL 21
These failures are:
two cases involving log2 and complex. My system returns NaN when
octave was expecting Inf.
several containing NaN and/or Inf in sort. In some cases my
system shows Nan when octave was expecting Inf.
I was almost sure as cygwin have the same discrepancy and good chunck
of cygwin math lib is taken by BSD...
several containing differences in numerical values which were within
2*eps.
propose a patch / change of that tolerances to avoid these false positive.
One using datevec. datevec("09/13") returns the year as 2000, not
the expected 2012.
One using getpwent(); Repeating endpwent(), x=getpwent() does not
return the same value for x, but, the next entry in passwd.
Knowing the Nan, Inf problem, these failures are not a serious detriment
to using octave 3.6.1 on my system.
Tom Dean
Marco