1 |
Mike Edenfield wrote: |
2 |
> On 6/25/2011 8:04 AM, Dale wrote: |
3 |
> |
4 |
>>> We restructured the dependency chain for fortran support, |
5 |
>>> which includes |
6 |
>>> a compile test now. The failure can be seen above. |
7 |
>>> |
8 |
>>> The Problem was in short, USE=fortran was enabled by |
9 |
>>> default for linux |
10 |
>>> arches, but people tend to disable it. Depending on |
11 |
>>> gcc[fortran] doesn't |
12 |
>>> work completely as gcc:4.4[fortran] and gcc:4.5[-fortran] |
13 |
>>> with gcc-4.5 |
14 |
>>> select can be installed, which would full fill the |
15 |
>>> dependency but |
16 |
>>> nevertheless doesn't give a working compiler. |
17 |
>>> |
18 |
>>> So now packages depend on virtual/fortran and use an |
19 |
>>> eclass to check for |
20 |
>>> a working compiler. So if you see this message, this means |
21 |
>>> you somehow |
22 |
>>> worked around gcc[fortran]. |
23 |
> |
24 |
>> That make sense? |
25 |
> |
26 |
> Yes. He's saying they didn't change the USE flag, they changed the |
27 |
> fortran dependency test to actually do a run-time check for fortran |
28 |
> because the USE flag alone wasn't sufficient. |
29 |
> |
30 |
> Which means you most likely had a non-working cantor and no fortran |
31 |
> compiler before and just didn't notice :) |
32 |
> |
33 |
> --Mike |
34 |
> |
35 |
> |
36 |
|
37 |
My understanding, USE flag was there and had been for a long time, got |
38 |
changed, this thread was started, discussion was had, USE flag was put |
39 |
back the way it was. So actually it was only not working while I was |
40 |
messing with it. That would be true ONLY if you were using the |
41 |
defaults. If you had -fortran then nothing should have changed as would |
42 |
having fortran enabled. It was only folks like me that didn't have any |
43 |
mention of fortran that were affected. |
44 |
|
45 |
Just one of those things. ;-) As someone else posted, this was minor |
46 |
compared to some things we have ran into. |
47 |
|
48 |
Dale |
49 |
|
50 |
:-) :-) |