Bug 157410 - sci-libs/lapack-reference-3.1.0.ebuild (version bump)
|
Bug#:
157410
|
Product: Gentoo Linux
|
Version: 2006.1
|
Platform: All
|
|
OS/Version: Linux
|
Status: RESOLVED
|
Severity: enhancement
|
Priority: P2
|
|
Resolution: FIXED
|
Assigned To: sci@gentoo.org
|
Reported By: ediap@users.sourceforge.net
|
|
Component: Ebuilds
|
|
|
URL:
http://www.netlib.org/lapack/index.html
|
|
Summary: sci-libs/lapack-reference-3.1.0.ebuild (version bump)
|
|
Keywords:
|
|
Status Whiteboard:
|
|
Opened: 2006-12-07 01:04 0000
|
Hi!
Please find the attached autotool patch for a new LAPACK (version 3.1.0) and a
patch with differences against lapack-reference-3.0-r4.ebuild.
BR,
/ediap
(In reply to comment #3)
> Created an attachment (id=103525) [edit] [details]
> lapack-reference-3.1.0-autotool.patch
>
> Fixed a problem with slamch.f and dlamch.f compilation without FFLAGS.
Eh? They're supposed to ignore FFLAGS, otherwise you get miscompilation.
(In reply to comment #4)
> > Fixed a problem with slamch.f and dlamch.f compilation without FFLAGS.
>
> Eh? They're supposed to ignore FFLAGS, otherwise you get miscompilation.
You are right. I just forgot about these two files while preparing a new patch,
since they are in a different directory now (INSTALL instead of SRC). Therefore
I posted an updated patch. In this new LAPACK release, there are also another
two files, which need to be compiled without any FFLAGS. Everything is in the
relevant Makefile.am.
BR,
/ediap
(In reply to comment #5)>
> You are right. I just forgot about these two files while preparing a new patch,
> since they are in a different directory now (INSTALL instead of SRC). Therefore
> I posted an updated patch. In this new LAPACK release, there are also another
> two files, which need to be compiled without any FFLAGS. Everything is in the
> relevant Makefile.am.
>
> BR,
> /ediap
>
Hi Adam,
Thanks much for your ebuild and the patches. I've just committed a
lapack-3.1.0 ebuild to portage cvs. In addition to your changes I have
- enabled lapack's test routines, such that now src_test() should run
through the full test set.
- changed the license from "lapack" to "BSD" since the tarball ships
with a COPYING file which is verbatim the BSD license.
I've currently package.mask'ed it for a few days of more testing. For me all
lapack test routines do pass as do itpp's. It would be great if you could
check it out as well if you have time sometime. Once it's unmasked
I'll bump lapack-atlas as well.
Thanks,
Markus
(In reply to comment #6)
[...]
> I've currently package.mask'ed it for a few days of more testing. For me all
> lapack test routines do pass as do itpp's. It would be great if you could
> check it out as well if you have time sometime. Once it's unmasked
> I'll bump lapack-atlas as well.
Hi Markus! I have just installed the lapack-reference-3.1.0 from official
portage and tested the IT++ library against it. No problems observed on my x86
(Intel Pentium-M). All tests passed.
BR,
/Adam
Hi Adam,
Thanks much for testing! I'll unmask 3.1.0 later today and will
hopefully be able to bump lapack-atlas as well.
Thanks,
Markus