[petsc-dev] "modules" and PETSc configure/make

[Satish Balay]

For one - we need to understand how modules are causing this
problem. This is an OS file - and should be in PATH. misusing modules
should not make a difference [wrt finding /usr/bin/printf]

>>>>>>>>
LOADEDMODULES=slurm/15.08.8:ppplcluster/1.2:subversion/1.8.11:python/2.7.2:bison/2.7:gcc4/4.8.1:openmpi/1.4.3:lapack-blas/3.3.1:git/1.8.0.2:swig/3.0.5:cmake/3.6.2

PATH=/p/fenics/fenics/local/bin:/p/nimrod/dbrennan/nimall/nimdevel/scripts:/p/nimrod/dbrennan/nimall-build/nimdevel/bin:/p/nimrod/dbrennan/nimall-tarball/autotools-lt-2.4.2/bin:/p/nimrod/dbrennan/nimall-tarball/xz-5.0.3-ser/bin:/p/nimrod/dbrennan/nimall-tarball/hdf5-1.8.13-ser/bin:/p/nimrod/dbrennan/nimall-tarball/cmake-3.4.1-ser/bin:/p/nimrod/dbrennan/nimall-tarball/python/bin:/p/nimrod/dbrennan/nimall-tarball/texlive/bin/x86_64-linux:/p/nimrod/dbrennan/nimall-build/nimdevel/ser/bin:/p/nimrod/dbrennan/nimall-tarball/bin:/p/nimrod/dbrennan/nimall/xdraw:/p/fenics/fenics/local/bin:/usr/pppl/cmake/3.6.2/bin:/usr/pppl/swig/3.0.5/bin:/usr/pppl/git/1.8.0.2/bin:/p/nimrod/dbrennan/nimall/nimdevel/scripts:/p/nimrod/dbrennan/nimall-build/nimdevel/bin:/p/nimrod/dbrennan/nimall-tarball/autotools-lt-2.4.2/bin:/p/nimrod/dbrennan/nimall-tarball/xz-5.0.3-ser/bin:/p/nimrod/dbrennan/nimall-tarball/hdf5-1.8.13-ser/bin:/p/nimrod/dbrennan/nimall-tarball/cmake-3.4.1-ser/bin:/p/nimrod/dbrennan/nimall-tarball/python/bin:/p/nimrod/dbrennan/nimall-tarball/texlive/bin/x86_64-linux:/p/nimrod/dbrennan/nimall-build/nimdevel/ser/bin:/p/nimrod/dbrennan/nimall-tarball/bin:/p/nimrod/dbrennan/nimall/xdraw:/usr/pppl/gcc/4.6-pkgs/openmpi-1.4.3/bin:/usr/pppl/gcc/4.8.1/bin:/usr/pppl/bison/2.7/bin:/usr/pppl/python/2.7.2/bin:/usr/pppl/subversion/1.8.11/bin:/usr/lib64/qt-3.3/bin:/usr/pppl/ppplcluster/1.2/bin:/usr/pppl/ppplcluster/1.2/sbin:/usr/pppl/slurm/15.08.8/bin:/usr/pppl/bin:/bin:/usr/bin:.:/u/dbrennan/bin:.
<<<<<<<

/usr/bin is in PATH. [if it were not - even 'ls' wouldn't work]

Or perhaps the script is looking for a file 'printf %s\n' - that
didn't work? But configure is invoking /bin/sh - and that should
always work..

BTW: This is a pppl machine :)

Satish

On Thu, 5 Oct 2017, Barry Smith wrote:

> 
>   Fuck "modules" to hell for all time!
> 
>   Is there a way to put some "module" sanity checks at the beginning of PETSc's configure?
> 
>    It seems everyone who touches a Nersc machine has module problems. We waste way to much time on "missing module" nonsense.
> 
>   Barry
> 
> 
> 
> > On Oct 3, 2017, at 10:39 PM, Mark Adams <mfadams at lbl.gov> wrote:
> > 
> > 
> > 
> > On Tue, Oct 3, 2017 at 4:22 PM, Satish Balay <balay at mcs.anl.gov> wrote:
> > >>>>>
> > checking size of bool... 0./configure: line 12435: printf %s\n: command not found
> > ./configure: line 12436: printf %s\n: command not found
> > configure: WARNING: The C++ compiler g++ cannot compile a program containing the <string> header - this may indicate a problem with the C++ installation.  Consider configuing with --disable-cxx
> > configure: WARNING: Structures containing long doubles may be aligned differently from structures with floats or longs.  MPICH does not handle this case automatically and you should avoid assumed extents for structures containing float types.
> > configure: error: unable to determine matching C type for C++ bool
> > <<<<<<
> > 
> > For some reason MPICH configure is failing.
> > 
> > 
> > balay at asterix /home/balay/petsc (maint=)
> > $ which printf
> > /usr/bin/printf
> > balay at asterix /home/balay/petsc (maint=)
> > $ rpm -qf /usr/bin/printf
> > coreutils-8.27-16.fc27.x86_64
> > balay at asterix /home/balay/petsc (maint=)
> > $
> > 
> > Your machine does not have this basic coreuitls package installed?
> > 
> > 
> > We are probably missing a module,
> >  
> > Satish
> > 
> > On Tue, 3 Oct 2017, Dylan P. Brennan wrote:
> > 
> > >
> > > Hello,
> > >
> > > I’m having problems configuring, any ideas?
> > >
> > > Dylan
> > >
> > >
> > 
> 
>