1 |
On 13 Feb, Ian Lee wrote: |
2 |
> Nikos Chantziaras wrote: |
3 |
>> Helmut Jarausch wrote: |
4 |
>>> On 13 Feb, Nikos Chantziaras wrote: |
5 |
>>>> Helmut Jarausch wrote: |
6 |
>>>>> Hi, |
7 |
>>>>> |
8 |
>>>>> I'm working on an AMD opteron, definitely an AMD64 machine. |
9 |
>>>>> |
10 |
>>>>> /etc/make.conf contains |
11 |
>>>>> CHOST="x86_64-pc-linux-gnu" |
12 |
>>>>> |
13 |
>>>>> my profile is |
14 |
>>>>> default/linux/amd64/2008.0/desktop |
15 |
>>>>> |
16 |
>>>>> Still, when emerging sci-libs/acml, portage |
17 |
>>>>> fetches the 32-bit variant of the package. |
18 |
>>>>> The SRC_URI depends on use flags x86 and amd86. |
19 |
>>>> This package is fetch restricted. It shouldn't actually fetch |
20 |
>>>> anything. You have to download the correct tarball and place it in |
21 |
>>>> /usr/portage/distfiles. |
22 |
>>>> |
23 |
>>> |
24 |
>>> OK, my wording was wrong. I did fetch the 64-bit version but emerge |
25 |
>>> didn't use it, but told me to fetch the 32-bit version. |
26 |
>> |
27 |
>> Hmm. Get that one too then and see which one is picked when the actual |
28 |
>> emerge starts. It could be that it needs both. If it ends up building |
29 |
>> the 32-bit one only, then I guess it's time for a bug report :) |
30 |
>> |
31 |
>> |
32 |
> |
33 |
> Its definitely a 64/32 bit problem and nothing to do with the int64 use |
34 |
> flag? |
35 |
> |
36 |
|
37 |
Sorry, it's probably (partially) my fault. |
38 |
I've tried to generate an overlay for version 4.2.0 (currently portage |
39 |
has only 4.1.0-rc1). |
40 |
Now, ebuild ... digest required to download all six packages ( 3 for gfortran |
41 |
plus 3 for ifort ). |
42 |
|
43 |
Sorry for the noise, |
44 |
Helmut. |
45 |
|
46 |
|
47 |
-- |
48 |
Helmut Jarausch |
49 |
|
50 |
Lehrstuhl fuer Numerische Mathematik |
51 |
RWTH - Aachen University |
52 |
D 52056 Aachen, Germany |