AMBER Archive (2007)

Subject: Re: AMBER: Problems during compilation and test.

From: David A. Case (case_at_scripps.edu)
Date: Tue Apr 10 2007 - 20:27:39 CDT


On Mon, Apr 09, 2007, Liang, Lei wrote:

> >> ./configure compaqf90

This is a compiler that not many people have, and you are seeing some the
things that can go wrong.

It's a little hard to read your latest e-mails, since you don't strip out
old comments. Some of the things listed look like they have already been
fixed(?)

> >>cc -c -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -O -std1 -o analyze.o
> >>analyze.c
> >>cc: Error: analyze.c, line 470: Invalid statement. (badstmt)
> >> eigval = (double *) safe_malloc(nelem * sizeof(double)); //
> >>for the eigenvalues
> >>---------------------------------------------------------------------^
> >>
> >
> >The C preprocessor on your platform is not stripping C++ style comments.

If you have not already done so, remove the comments from analyze.c (line
470). ANSI C99 supports what Amber has, so you are trying to compile with
quite an old C compiler.

> We modified the config.h file as following at the very end: ...
> otherwise it has difficulty to recognize the 'amoeba_parm.f90' as the
> input file for the c preprocessor.
> >>Now the compilation is successful.

Sounds like this now works...I'm very confused about why this is needed,
since the rule for compiling amoeba_parm.f90 is given explicitly in the
Makefile -- there should be no need for a general rule. But, since it now
works, lets move on.

>
> cd water2_qmmmnmr; ./Run.wnmr
> forrtl: error (65): floating invalid

If this is only error, you are in (pretty) good shape: unless you are
planning to use the "NMR" feature of QM/MM, which is rare, you can probably
afford to ignore this. Anyway, put this aside, and let's see if there are
any other problems that affect key functionality. We can work on the corner
cases later.

[Of course, any one with access to a compaq f90 compiler is welcome to help
debug some of these problems; they are not all that unusual, given that the
compilers (and even the version of make) might be quite different from what
most people have tested Amber on.]

...good luck...dac
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu