[NumPy-Tickets] [NumPy] #1322: Scipy build fails on Mac OS X 10.5.8, Numpy import fails on Mac OS X 10.5.8

NumPy Trac numpy-tickets@scipy....
Tue Sep 7 08:52:45 CDT 2010

#1322: Scipy build fails on Mac OS X 10.5.8, Numpy import fails on Mac OS X 10.5.8
 Reporter:  ndufour                                                |       Owner:  cdavid
     Type:  defect                                                 |      Status:  new   
 Priority:  high                                                   |   Milestone:  1.5.1 
Component:  Official binaries                                      |     Version:        
 Keywords:  build failure import failure multiarray.so multiarray  |  
Changes (by rgommers):

  * status:  needs_info => new


 Reformatting the first comment for readability:

 packages/numpy/core/multiarray.so: Mach-O universal binary with 2
 packages/numpy/core/multiarray.so (for architecture ppc7400): Mach-O
 bundle ppc
 packages/numpy/core/multiarray.so (for architecture i386): Mach-O bundle

 /Library/WebServer/CGI-Executables>uname -a
 Darwin dhcp-18-111-92-239.dyn.mit.edu 9.8.0 Darwin Kernel Version 9.8.0:
 Wed Jul 15 16:55:01 PDT 2009; root:xnu-1228.15.4~1/RELEASE_I386 i386

 I've looked into this some more and found a large amount of reports with
 the same problem - no solution. The issue seems to that binaries compiled
 on 10.6 simply fail to load on 10.5 despite the correct SDK being used
 (10.4 in this case, I checked all individual files in the build log). I
 think this problem first started occurring when David C. upgraded his
 machine to Snow Leopard before 1.4.0.

 One of the most complete threads about this I found is
 http://web.archiveorange.com/archive/v/JT06FFoTBH5HKDgJ9T5X. At the end
 there is a report that including a compile flag {{{-mmacosx-version-
 min=10.4}}} helps. Will give that a try when I find the time.

 The only safe way to go though, as far as I can tell, is what Sage does:
 provide separate binaries for each version of OS X, built on that same OS

Ticket URL: <http://projects.scipy.org/numpy/ticket/1322#comment:14>
NumPy <http://projects.scipy.org/numpy>
My example project

More information about the NumPy-Tickets mailing list