[Pw_forum] Linux ia32 required additional flags and emits warnings

[Axel Kohlmeyer]

On Tue, 10 Apr 2007, Alexander McPhail wrote:


dear vivian,

(first off, please configure your mailer so that it shows the 
correct name. it is very confusing to start writing a reply to
alexander and then see you're replying to vivian. thanks.).

VM> Hi,
VM> 
VM> When trying to compile the complete espresso distribution with patch level
VM> 3.2.1 I found that I had to include a couple of extra flags in the configure
VM> command
VM> 
VM> IFLAGS=
VM> -I../include : so that in the compilation of iotk_error.f90 include files
VM> were found
VM> -I/usr/include/mpi : otherwise these headers were not found, even though
VM> libmpi was found and configured

hmmm. that looks a lot like your MPI library is not 
correctly configured. the configure script has actually 
very little means to correctly detect a working fortran
parallel programming environment (configure was mostly written
to detect the (smallish) differences in unix installations).
this is much more complicated on a linux machine because there
are so many ways that linux machines can be set up or installed,
not to mention the multitude of different fortran (and c) compilers.
the problem with fortran is, that - in general - you cannot mix
and match different compilers. so you have to make sure that your
whole toolchain is compiled and set up consistently. configure
only contains a number of 'educated guesses'. if you want to compile
QE in parallel, you almost always have to adapt the resulting 
make.sys, in some cases even rewrite in large parts (see the files
in the install directory that are templates for machines, where 
configure runs into big troubles or fails completely).

VM> Also, during compilation there were many warnings like:
VM> 
VM> In file cegterg.f90:349
VM> <snip>
VM> Warning (155): Inconsistent types (COMPLEX(8)/REAL(8)) in actual argument
VM> lists at (1) and (2)
VM> 
VM> Are these a problem?

g95 is a very picky compiler. it is quite common for fortran program
to alias complex arrays to real arrays.

cheers,
  axel.

VM> 
VM> Thanks,
VM> 
VM> Vivian
VM> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.