From: ageay <ageay>
Date: Thu, 18 Jul 2013 13:15:09 +0000 (+0000)
Subject: On MD10 32 bits OS numpy bug on 1.6.2 on cumsum ???
X-Git-Tag: B4KillOfAutomake~1
X-Git-Url: http://git.salome-platform.org/gitweb/?a=commitdiff_plain;h=b1bb8dfd2ca2e09e5893bd814f3da9424fa92449;p=modules%2Fmed.git

On MD10 32 bits OS numpy bug on 1.6.2 on cumsum ???
Adrien is on this !
For the moment on 32bit architecture the test is skipped !
---

diff --git a/src/MEDCoupling_Swig/MEDCouplingNumPyTest.py b/src/MEDCoupling_Swig/MEDCouplingNumPyTest.py
index 78dc02d18..0fcfe6c3f 100644
--- a/src/MEDCoupling_Swig/MEDCouplingNumPyTest.py
+++ b/src/MEDCoupling_Swig/MEDCouplingNumPyTest.py
@@ -25,13 +25,14 @@ if MEDCouplingHasNumPyBindings():
     from numpy import *
     pass
 
+from platform import architecture
 from sys import getrefcount
 
 import os,gc,weakref,unittest
 
 class MEDCouplingNumPyTest(unittest.TestCase):
     
-    @unittest.skipUnless(MEDCouplingHasNumPyBindings(),"requires numpy")
+    @unittest.skipUnless(MEDCouplingHasNumPyBindings() and architecture()[0]=="64bit","requires numpy")
     def test1(self):
         sz=20
         a=array(0,dtype=int32)