From f304b48296f679e5605210fc19f36fa43e569927 Mon Sep 17 00:00:00 2001 From: Marshall Farrier Date: Fri, 9 Jan 2015 18:28:23 -0800 Subject: DOC: Fix doc of seterrcall. Issue #4427 --- numpy/core/numeric.py | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) (limited to 'numpy/core/numeric.py') diff --git a/numpy/core/numeric.py b/numpy/core/numeric.py index 1847300dd..c1c555172 100644 --- a/numpy/core/numeric.py +++ b/numpy/core/numeric.py @@ -2675,12 +2675,13 @@ def seterrcall(func): Function to call upon floating-point errors ('call'-mode) or object whose 'write' method is used to log such message ('log'-mode). - The call function takes two arguments. The first is the - type of error (one of "divide", "over", "under", or "invalid"), - and the second is the status flag. The flag is a byte, whose - least-significant bits indicate the status:: + The call function takes two arguments. The first is a string describing the + type of error (such as "divide by zero", "overflow", "underflow", or "invalid value"), + and the second is the status flag. The flag is a byte, whose four + least-significant bits indicate the type of error, one of "divide", "over", + "under", "invalid":: - [0 0 0 0 invalid over under invalid] + [0 0 0 0 divide over under invalid] In other words, ``flags = divide + 2*over + 4*under + 8*invalid``. -- cgit v1.2.1