1 |
EL Thu, 2 Jun 2011 07:22:18 +1200 |
2 |
Francois Bissey <fbissey@××××××××××××.nz> escrigué: |
3 |
|
4 |
> > EL Wed, 1 Jun 2011 06:57:03 +1200 |
5 |
> > |
6 |
> > Francois Bissey <fbissey@××××××××××××.nz> escrigué: |
7 |
> > > > Hi, |
8 |
> > > > |
9 |
> > > > While updating sage and python to 4.7 and 2.7, I hit a |
10 |
> > > > variation of an old bug (see http://bugs.gentoo.org/195619): |
11 |
> > > > scipy compiles with lapack-reference but is unusable. Running |
12 |
> > > > in python |
13 |
> > > > |
14 |
> > > > import scipy.optimize |
15 |
> > > > |
16 |
> > > > fails with |
17 |
> > > > |
18 |
> > > > [...] |
19 |
> > > > |
20 |
> > > > from scipy.linalg import clapack |
21 |
> > > > |
22 |
> > > > ImportError: |
23 |
> > > > /usr/lib64/python2.7/site-packages/scipy/linalg/clapack.so: |
24 |
> > > > undefined symbol: clapack_sgesv |
25 |
> > > > |
26 |
> > > > I tried rebuilding numpy and scipy with lapack provided |
27 |
> > > > (eselect) by lapack-reference, blas provided by blas-reference, |
28 |
> > > > cblas provided by gsl and with/without clapack with the same |
29 |
> > > > result. |
30 |
> > > > |
31 |
> > > > Then I tried rebuilding numpy and scipy with lapack, blas and |
32 |
> > > > cblas provided (eselect) by atlas and scipy worked flawlessly. |
33 |
> > > > |
34 |
> > > > Should scipy depend on a specific lapack provider instead of a |
35 |
> > > > virtual? Enable it at compile time? Is it an unfortunate |
36 |
> > > > combination what fails? |
37 |
> > > > |
38 |
> > > > I can provide build logs if you need them. |
39 |
> > > |
40 |
> > > Hi Marc, |
41 |
> > > |
42 |
> > > Just one thing that I can think of on the top of my head. We |
43 |
> > > recently made scipy depend on arpack and superlu. They also have |
44 |
> > > dependencies on blas/lapack if you didn't rebuild them as well |
45 |
> > > they may cause trouble. |
46 |
> > > |
47 |
> > > I had a similar problem recently but eselecting everything |
48 |
> > > properly worked for me. A last piece of info you could provide: |
49 |
> > > are you using the main tree implementation or bicatali's overlay |
50 |
> > > implementation? |
51 |
> > > |
52 |
> > > Francois |
53 |
> > |
54 |
> > Hi Francois, |
55 |
> > |
56 |
> > I tried rebuilding arpack and superlu without luck. Since you |
57 |
> > mentioned that eselect fixed your problems, I also tried installing |
58 |
> > cblas-reference and using it instead of gsl, also without luck. |
59 |
> > |
60 |
> > I'm using the main tree implementation. |
61 |
> > |
62 |
> Hi Marc, |
63 |
> |
64 |
> I have been using bicatali's overlay for a bit and I recompiled all |
65 |
> the cblas/lapack stack from sage a few times with different |
66 |
> implementations in the last few days. The only time i had an issue |
67 |
> was when the value eselected was not what I wanted by accident. |
68 |
> |
69 |
> I am curious to see your build log for numpy and scipy. Also what |
70 |
> does ldd /usr/lib64/python2.7/site-packages/scipy/linalg/clapack.so |
71 |
> reports? |
72 |
> |
73 |
> Francois |
74 |
> |
75 |
|
76 |
Here are my logs. You'll find some warnings in the configuration |
77 |
stage, I hope they'll mean something to you. Also, there's no trace of |
78 |
libclapack.so in |
79 |
ldd /usr/lib64/python2.7/site-packages/scipy/linalg/clapack.so. |