Re: AMBER: building AMBER9 with gfortran/cygwin

From: Andreas Svrcek-Seiler <svrci.tbi.univie.ac.at>
Date: Fri, 25 Aug 2006 12:39:08 +0200 (CEST)

Hi,
> I am trying to build AMBER9 on CYGWIN with gcc 4.1.1, most modules can be
> compiled successfully, however, it failed at compiling sander:
>
> character(len=len(MVPM_FORWARD)) ::
> matrix_vector_product_method=MVPM_FORWARD
> 1
> Error: 'matrix_vector_product_method' at (1) must have constant character
> length in this context
> In file _lmod.f:262
>
> character(len=len(MC_METHOD_QUICK_QUENCH)) :: Monte_Carlo_method &
> 1
> Error: 'monte_carlo_method' at (1) must have constant character length in
> this context
> In file _lmod.f:325
>
> character(len=len(XMIN_METHOD_LBFGS)) :: xmin_method = XMIN_METHOD_LBFGS
                                                     1
> Error: 'xmin_method' at (1) must have constant character length in this

Here's a workaround:

182c182
< character(len=*) :: matrix_vector_product_method=MVPM_FORWARD 
---
>    character(len=len(MVPM_FORWARD)) :: 
matrix_vector_product_method=MVPM_FORWARD
194c194
<    character(len=*) :: Monte_Carlo_method &
---
>    character(len=len(MC_METHOD_QUICK_QUENCH)) :: Monte_Carlo_method &
257c257
<    character(len=*) :: xmin_method = XMIN_METHOD_LBFGS
---
>    character(len=len(XMIN_METHOD_LBFGS)) :: xmin_method = XMIN_METHOD_LBFGS
...just change lines from the diff above in lmod.f - then compilation runs 
through (at least here). Note that
1) I'm not sure the lmod routines work after applying this patch.
2) This seems to be a workaround for a compiler-bug (not sure-
my F90 knowledge is quite limited).
3) make test.sander runs through except
for the noesy test, which terminates with:
Namelist reports error in reading noeexp
  -- Subscript out of range implies dimensioning problem
  -- (see nmr.h)
...which seems unrelated to the patch above
4) (related to point 2): ifort doesn't compile lmod.f after 
applying this patch.
I hope this helps,
good luck
Andreas
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Aug 27 2006 - 06:07:22 PDT
Custom Search