1 |
On 22/06/11 06:31, Matthew Finkel wrote: |
2 |
> On 06/21/11 23:55, Dale wrote: |
3 |
>> I just did my updates and ran into this: |
4 |
>> |
5 |
>> * Maintainer: sci@g.o |
6 |
>> * USE: amd64 consolekit elibc_glibc kernel_linux multilib |
7 |
>> policykit userland_GNU |
8 |
>> * FEATURES: preserve-libs sandbox |
9 |
>> |
10 |
>> * Please install currently selected gcc version with USE=fortran. |
11 |
>> * If you intend to use a different compiler then gfortran, please |
12 |
>> * set FC variable accordingly and take care that the neccessary |
13 |
>> * fortran dialects are support. |
14 |
>> |
15 |
>> * ERROR: sci-libs/blas-reference-20070226 failed (setup phase): |
16 |
>> * Currently no working fortran compiler is available |
17 |
>> * |
18 |
>> * Call stack: |
19 |
>> * ebuild.sh, line 56: Called pkg_setup |
20 |
>> * ebuild.sh, line 1446: Called fortran-2_pkg_setup |
21 |
>> * fortran-2.eclass, line 134: Called _die_msg |
22 |
>> * fortran-2.eclass, line 120: Called die |
23 |
>> * The specific snippet of code: |
24 |
>> * die "Currently no working fortran compiler is available" |
25 |
>> * |
26 |
>> * If you need support, post the output of 'emerge --info |
27 |
>> =sci-libs/blas-reference-20070226', |
28 |
>> * the complete build log and the output of 'emerge -pqv |
29 |
>> =sci-libs/blas-reference-20070226'. |
30 |
>> * The complete build log is located at |
31 |
>> '/var/log/portage/sci-libs:blas-reference-20070226:20110622-034357.log'. |
32 |
>> * The ebuild environment file is located at |
33 |
>> '/var/tmp/portage/sci-libs/blas-reference-20070226/temp/die.env'. |
34 |
>> * S: |
35 |
>> '/var/tmp/portage/sci-libs/blas-reference-20070226/work/lapack-lite-3.1.1' |
36 |
>> |
37 |
>>>>> Failed to emerge sci-libs/blas-reference-20070226, Log file: |
38 |
>> |
39 |
>>>>> |
40 |
>> '/var/log/portage/sci-libs:blas-reference-20070226:20110622-034357.log' |
41 |
>> root@fireball / # |
42 |
>> |
43 |
>> This is my gcc info: |
44 |
>> |
45 |
>> [ebuild R ] sys-devel/gcc-4.4.5 USE="gtk mudflap (multilib) nls |
46 |
>> nptl openmp (-altivec) -bootstrap -build -doc (-fixed-point) -fortran |
47 |
>> -gcj -graphite (-hardened) (-libffi) -multislot -nocxx -nopie -nossp |
48 |
>> -objc -objc++ -objc-gc -test -vanilla" |
49 |
>> |
50 |
>> So, does everyone need to turn on the fortran USE flag so that they |
51 |
>> don't break anything? May I also add, the USE flag description is |
52 |
>> worth about as much as a screen door on a submarine. |
53 |
>> |
54 |
>> fortran - Adds support for fortran (formerly f77) |
55 |
>> |
56 |
>> That doesn't tell me very much. |
57 |
>> |
58 |
>> Heads up for folks about to do their updates, check into the USE flag |
59 |
>> fortran to see if you need to add it to yours before updating a bunch |
60 |
>> of stuff. |
61 |
>> |
62 |
>> Dale |
63 |
>> |
64 |
>> :-) :-) |
65 |
>> |
66 |
> If I had to guess, I'd say =sci-libs/blas-reference-20070226 requires |
67 |
> fortran (ebuild depends on it) and you don't have another fortran |
68 |
> compiler installed. |
69 |
> |
70 |
> Could be wrong though. |
71 |
> |
72 |
> - Matt |
73 |
> |
74 |
|
75 |
That's right, |
76 |
|
77 |
blas-reference is written in fortran. |
78 |
|
79 |
We restructured the dependency chain for fortran support, which includes |
80 |
a compile test now. The failure can be seen above. |
81 |
|
82 |
The Problem was in short, USE=fortran was enabled by default for linux |
83 |
arches, but people tend to disable it. Depending on gcc[fortran] doesn't |
84 |
work completely as gcc:4.4[fortran] and gcc:4.5[-fortran] with gcc-4.5 |
85 |
select can be installed, which would full fill the dependency but |
86 |
nevertheless doesn't give a working compiler. |
87 |
|
88 |
So now packages depend on virtual/fortran and use an eclass to check for |
89 |
a working compiler. So if you see this message, this means you somehow |
90 |
worked around gcc[fortran]. |
91 |
|
92 |
|
93 |
justin |