[SciPy-dev] Stupid question.
eric at scipy.org
Tue Feb 19 13:05:25 CST 2002
> On Tue, 19 Feb 2002, Pearu Peterson wrote:
> > See fftw libraries section in
> > http://www.scipy.org/site_content/tutorials/build_instructions
> Already done that. I'm using exactly the rpms from that very link.
> Should I build them from scratch? I don't see why that should make any
More than that, it was working before right? From the command line link
statement you sent, it looks like it *should* link. If you type it at the
command line does it work? Maybe add -L/usr/local/lib and see if that helps?
The test checkin last night shouldn't have affected this. I did make some other
changes to build_ext.py because scipy_distutils wasn't picking up 'g2c' etc.
when linking in Fortran libraries. Perhaps this did something bad on Linux, but
I wouldn't think so.
Here is the change I made:
if linker_so is not None:
if linker_so is not save_linker_so:
print 'replacing linker_so %s with %s'
self.compiler.linker_so = linker_so
l = build_flib.get_fcompiler_library_names()
#l = self.compiler.libraries + l
self.compiler.libraries = l
l = build_flib.get_fcompiler_library_dirs()
#l = self.compiler.library_dirs + l
self.compiler.library_dirs = l
# I added this else statment and its contents
libs = build_flib.get_fcompiler_library_names()
for lib in libs:
if lib not in self.compiler.libraries:
#self.compiler.libraries = self.compiler.libraries + l
lib_dirs = build_flib.get_fcompiler_library_dirs()
for lib_dir in lib_dirs:
if lib_dir not in self.compiler.library_dirs:
#self.compiler.library_dirs = self.compiler.library_dirs + l
More information about the Scipy-dev