[Numpy-tickets] [NumPy] #539: config parameters are not passed on to build

NumPy numpy-tickets@scipy....
Tue Jun 26 17:14:40 CDT 2007


#539: config parameters are not passed on to build
---------------------------------+------------------------------------------
 Reporter:  pv                   |        Owner:  cookedm
     Type:  defect               |       Status:  new    
 Priority:  normal               |    Milestone:         
Component:  numpy.distutils      |      Version:         
 Severity:  normal               |   Resolution:         
 Keywords:  distutils config -I  |  
---------------------------------+------------------------------------------
Comment (by rkern):

 The options on the config command are not supposed to propagate to the
 build_ext command. They exist to configure the compiler that is used
 during the config stage when we compile and link some small C programs in
 order to probe the system for its capabilities (namely long doubles, some
 of the C99 special functions, and the version of ATLAS).

 All of the build_ext options that you say don't work do work for me with
 your example and numpy 1.0.3 and r3875.

 build_clib does not need -L. It is possible that it should support -I, but
 it's not likely to be of much use.

 All of the --noarch, --noopt, and --debug options to config_fc do work for
 me using --fcompiler=gnu95 and numpy 1.0.3. They do not work for me with
 numpy r3875.

 Yes, you do need to specify --compiler multiple times. Put them all in
 your ~/.pydistutils.cfg, if you don't want to bother with them all the
 time.

 For the items that work for me, but not for you, please provide us the
 exact commands that you executed and the output you received. Otherwise,
 the only thing in this ticket to fix is the SVN regression with --noarch,
 etc. David, can you please take a look at this?

-- 
Ticket URL: <http://projects.scipy.org/scipy/numpy/ticket/539#comment:1>
NumPy <http://projects.scipy.org/scipy/numpy>
The fundamental package needed for scientific computing with Python.


More information about the Numpy-tickets mailing list