[Pkg-openmpi-maintainers] Bug#456721: Processed: Re: Bug#456721: libpetsc.so depends on unexistent libraries

[Adam C Powell IV]

On Mon, 2007-12-17 at 16:24 -0600, Dirk Eddelbuettel wrote:
> On 17 December 2007 at 22:27, Manuel Prinz wrote:
> | Am Montag, den 17.12.2007, 14:47 -0600 schrieb Dirk Eddelbuettel:
> | > On 17 December 2007 at 21:13, Manuel Prinz wrote:
> | No, as there are more libraries needed for the compiler wrappers, if I'm
> | not mistaken.
> | > I'm at work too so I didn't have time to check yet.  Recall that the final
> | > links are actually created by the debhelper tools in conjunction with ldd, ie
> | > what 'dpkg -c' typically shows at the end 
> | > 
> | > lrwxrwxrwx root/root         0 2007-12-12 13:13 ./usr/lib/openmpi/lib/libopen-pal.so.0 -> libopen-pal.so.0.0.0
> | > lrwxrwxrwx root/root         0 2007-12-12 13:13 ./usr/lib/openmpi/lib/libmpi.so.0 -> libmpi.so.0.0.0
> | > lrwxrwxrwx root/root         0 2007-12-12 13:13 ./usr/lib/openmpi/lib/libmpi_f77.so.0 -> libmpi_f77.so.0.0.0
> | > lrwxrwxrwx root/root         0 2007-12-12 13:13 ./usr/lib/openmpi/lib/libmpi_cxx.so.0 -> libmpi_cxx.so.0.0.0
> | > lrwxrwxrwx root/root         0 2007-12-12 13:13 ./usr/lib/openmpi/lib/libmpi_f90.so.0 -> libmpi_f90.so.0.0.0
> | > lrwxrwxrwx root/root         0 2007-12-12 13:13 ./usr/lib/openmpi/lib/libopen-rte.so.0 -> libopen-rte.so.0.0.0
> | > lrwxrwxrwx root/root         0 2007-12-12 13:13 ./usr/lib/openmpi/lib/libmca_common_sm.so.0 -> libmca_common_sm.so.0.0.0
> | > 
> | > are already symlinks.
> | 
> | This is not a problem, it's just useless. We need to
> | link /usr/lib/openmpi/lib/* to /usr/lib/ which should be done via
> | alternatives.
> 
> Are you sure we need alternatives for something like libmpi_cxx.so.0 which
> the 'other' (ie LAM) doesn't have?
> 
> AFAIK update-alternatives is for several implementations of the same file
> only.
> 
> Or are you in fact suggesting to use update-alternatives for libmpi.so _and_
> and the same time update all those libmpi_$Foo.so that only appear with Open
> MPI ?  That, I guess, would still be correct use of alternatives via the
> slave.  In fact there may or may not be exampel cases from the vi use or some
> other u-a use.

That already happens via alternatives slaves.  As discussed earlier,
it's inappropriate with ABI-incompatible soname-named files e.g. *.so.0

I think we're going in the right direction: alternatives for *.so and
different filenames in /usr/lib for *.so.0 , right?

The only way we could do better is to use the same names for our fortran
and C++ library .so alternatives.  But I don't think it's worth it: I
don't think they're even sufficiently API-compatible.

Cheers,
-Adam
-- 
GPG fingerprint: D54D 1AEE B11C CE9B A02B  C5DD 526F 01E8 564E E4B6

Engineering consulting with open source tools
http://www.opennovation.com/