Skip to Content.
Sympa Menu

cado-nfs - Re: [Cado-nfs-discuss] Fw:Re:Re: makefb not found in distributed enviroment

Subject: Discussion related to cado-nfs

List archive

Re: [Cado-nfs-discuss] Fw:Re:Re: makefb not found in distributed enviroment


Chronological Thread 
  • From: Emmanuel Thomé <emmanuel.thome@gmail.com>
  • To: meng <qsmeng@126.com>
  • Cc: "cado-nfs-discuss@lists.gforge.inria.fr" <cado-nfs-discuss@lists.gforge.inria.fr>
  • Subject: Re: [Cado-nfs-discuss] Fw:Re:Re: makefb not found in distributed enviroment
  • Date: Thu, 25 Jul 2013 15:56:11 +0200
  • List-archive: <http://lists.gforge.inria.fr/pipermail/cado-nfs-discuss>
  • List-id: A discussion list for Cado-NFS <cado-nfs-discuss.lists.gforge.inria.fr>

On Thu, Jul 25, 2013 at 2:46 PM, meng <qsmeng@126.com> wrote:
> With 8efeb35, the repeat problem still exists. With gdb, it displays

Just to make sure -- did you recompile ?

> I post my problem on openmpi in openmpi user list. LInux is free but
> troubling.

All systems, no matter which kind, require one to have basic
understanding of their mechanisms in order to compile and run
software.

On Thu, Jul 25, 2013 at 11:12 AM, meng <qsmeng@126.com> wrote:
> By the way, I am trying to install openmpi with the following problem.
> ./scripts/mpi_imrecv_f90.f90.sh
> /home/greatnet/openmpi-1.7.2/ompi/mpi/fortran/use-mpi-tkr >
> mpi_imrecv_f90.f90
> ../../../../libtool: line 1128: gfortran: command not found

when you mean "install", is the failure above something which happens
while you are compiling openmpi ? Please arrange so that others do not
have to rely on guesswork to fill in the blanks of your reports.

Please note also that 1.7.2 is a so-called "feature release", and you
might have better success with 1.6.5 instead (marked "current
release").

> However, I am sure that gfortran exists and can run in any directory.

If you so claim. But I would happily add that anyway, we do not care
at all about gfortran for cado-nfs usage.

[Openmpi's build process should detect the (un)availability of fortran
at compile time, and decide or not to compile the fortran bindings
accordingly. This automatic configuration choice carries over to all
test/check and install targets. If in spite of its inability to reach
gfortran, openmpi does fortran stuff, it's probably a bug. But again,
we don't care here].

openmpi's configure script has many arguments, two of which may be
specified to disable all fortran stuff. ./configure --help should
enlighten you.


> When I
> add "$PATH" after line 1128 in libtool, and then remake, it display
> "/usr/bin:/bin".

Never tinker with these scripts except in desperate circumstances.

> Yes, indeed my gcc is not in /usr/bin ,/bin.
> How to add
> gcc to that path or how to solve my problem?

Guess what ? The autotools do have provision for the user to specify a
custom compiler. And again, ./configure --help is the source of
information.

Be also aware that the autotools on your rhel5 machine are certainly
ice age old. If at some point you see automake, autoconf, or
libtoolize being run, then you are bound to encounter serious issues.
Adding --disable-maintainer-rules to the configure script sometimes
helps getting rid of this hazard, although I don't know if it's
actually used by openmpi or not.

E.





Archive powered by MHonArc 2.6.19+.

Top of Page