1 |
On 21/01/14 23:46, Sébastien Fabbro wrote: |
2 |
> Some applications (I can see armadillo, scamp) actually use the |
3 |
> atlas-specific routines clapack_* routines. |
4 |
> We could aim at keeping both the library soname and filename as close to |
5 |
> upstream, and add an option to the alternative framework to create a |
6 |
> ldscript with a generic soname and filename, turned on for blas and |
7 |
> cblas at least. |
8 |
|
9 |
|
10 |
I would really like to see that ldscript to be present. it would solve |
11 |
many propblems including linking in distutils against mkl. |
12 |
|
13 |
|
14 |
|
15 |
> |
16 |
> As a user, I don't particularly enjoy rebuilding octave, ROOT, R, or the |
17 |
> scipy stack. @preserved-rebuild is more a hack than a feature. |
18 |
> I don't know how many people actually switch providers, besides a first |
19 |
> benchmark test. My guess is probably not many given the burden of |
20 |
> re-compiling all the reverse dependencies and the previous fragility of |
21 |
> the system. |
22 |
|
23 |
But I still don't see the point of equalizing the soname. One advantage |
24 |
I see from separate sonames is that you can link different applications |
25 |
to different implementations. Of course this would require recompilation |
26 |
of revdeps in case you remove one from your system. But I would argue |
27 |
that switching/removing an implementation is a rare case for the average |
28 |
user. Most will pick one or take the default aka reference |
29 |
implementation and never touch it again. |
30 |
|
31 |
|
32 |
Justin |