| 1 |
Hi guys, |
| 2 |
|
| 3 |
I'm using freebsd-7.2 Release/i386. Bootstrapped successfully |
| 4 |
(gcc/gfortran built with USE="fortran") and I ran into problems |
| 5 |
emerging dev-python/numpy-1.3.0: |
| 6 |
|
| 7 |
$ emerge numpy-1.3.0 |
| 8 |
... |
| 9 |
build/src.freebsd-7.2-RELEASE-i386-2.6/numpy/core/src/umathmodule.c |
| 10 |
/usr/include/fenv.h: Assembler messages: |
| 11 |
/usr/include/fenv.h:156: Error: suffix or operands invalid for `fnstsw' |
| 12 |
---------------------------------------------------------------------------------------------------------- |
| 13 |
I think what's happening is that gcc is somehow pulling in |
| 14 |
/usr/include/fenv.h into the ebuild process. I can successfully build |
| 15 |
directly from the var/tmp/.../work directory: |
| 16 |
|
| 17 |
$ python setup.py build |
| 18 |
|
| 19 |
build/temp.freebsd-7.2-RELEASE-i386-2.6/build/src.freebsd-7.2-RELEASE-i386-2.6/numpy/core/src/umathmodule.o |
| 20 |
-Lbuild/temp.freebsd-7.2-RELEASE-i386-2.6 -lnpymath -lm -lpython2.6 -o |
| 21 |
build/lib.freebsd-7.2-RELEASE-i386-2.6/numpy/core/umath.so |
| 22 |
... |
| 23 |
|
| 24 |
So I don't know where in the ebuild the /usr/include/fenv.h is getting |
| 25 |
sucked in instead of telling python to use the |
| 26 |
numpy/core/include/numpy/fenv/fenv.h (which is newer) |
| 27 |
This may be related to the numpy issue on |
| 28 |
http://projects.scipy.org/numpy/ticket/811. |
| 29 |
|
| 30 |
A workaround is to copy the numpy/core/include/numpy/fenv/fenv.h to |
| 31 |
/usr/include/ as root but that's obviously cheating :) |
| 32 |
|
| 33 |
recommendations? |
Archives