Bug#612621: ERROR: SCOTCH_dgraphInit: linking with both libScotch and libPTScotch is not allowed

[Johannes Ring]

Hi Adam,

On Mon, Mar 28, 2011 at 3:55 PM, Adam C Powell IV <hazelsct at debian.org> wrote:
> Hi Johannes,
>
> On Mon, 2011-03-28 at 13:44 +0200, Johannes Ring wrote:
>> On Mon, Mar 28, 2011 at 12:56 PM, Pierre Saramito
>> <Pierre.Saramito at imag.fr> wrote:
>> > Hello !
>> >
>> > Despite the recent patch applied in version 5.1.11.dfgs-4,
>> > the bug seems to be still present on sid/amd64 :
>>
>> Yes, I also noticed that. I have reopened the bug now.
>>
>> >    uname -a
>> >        Linux mobydick 2.6.32-5-amd64 #1 SMP Wed Jan 12 03:40:32 UTC 2011 x86_64 GNU/Linux
>> >
>> >    apt-cache policy libptscotch-dev
>> >    libptscotch-dev:
>> >      Installé : 5.1.11.dfsg-4
>> >      Candidat : 5.1.11.dfsg-4
>> >     Table de version :
>> >     *** 5.1.11.dfsg-4 0
>> >        500 http://ftp.fr.debian.org/debian/ sid/main amd64 Packages
>> >        100 /var/lib/dpkg/status
>> >
>> >
>> >    mpic++ -o ptscotch_test ptscotch_test.cpp -DSCOTCH_PTSCOTCH -I/usr/include/scotch -lptscotch -lptscotcherr
>> >    ./ptscotch_test
>> >        (0): ERROR: SCOTCH_dgraphInit: linking with both libScotch and libPTScotch is not allowed
>> >        MPI implementation is not thread-safe:
>> >        SCOTCH should be compiled without SCOTCH_PTHREAD
>> >
>> > Yet, scotch library is unusable...
>> > Any idea to get free from this problem ?
>>
>> Yes, applying the attached patch to the git repository and rebuilding
>> the package fixes the problem for me.
>>
>> Adam: Thanks for your work on this package! Could you please
>> incorporate this fix in the next upload?
>
> Sure, thanks for the patch.  Oh wait -- I think the libscotch$(LIB)
> dependency and -lscotch linkage were needed for binutils-gold, have you
> tested the patch with binutils-gold?

How exactly do I do that? I tried to install the binutils-gold package
and then rebuilt the packages with the patch. That worked fine. Then I
installed these new packages and ran the ptscotch_test2.cpp, which
also worked fine. Is this correct?

> There's also a lex-related FTBFS bug (just filed) which I'll need to
> hunt down, and won't have time until the weekend or maybe next week, so
> it will be a little while before the next upload. :-(
>
> That said, if someone else fixes the FTBFS, I'll upload sooner.

I would like to help but I couldn't reproduce the FTBFS, unfortunately.

Johannes