[gmx-developers] Small configure.ac change

[Charlie Peck]

On Jan 1, 2005, at 5:15 AM, David van der Spoel wrote:

> On Fri, 31 Dec 2004, Charlie Peck wrote:
>
>> I'm working with the CVS snapshot labeled GROMACS 3.3_beta_20041226.
>> configure.ac produces a symbol GMX_MPI when --enable-mpi is specified
>> on the configure command line but many places in the source code are
>> looking for the symbol USE_MPI.  The net effect is that you can't run
>> this snapshot with MPI.
> I haven't seen this problem in plain CVS. Maybe you should just try to
> download the latest cvs rather than this snapshot.

I misspoke, I am using the latest CVS cut (retrieved with $ cvs ... co 
gmx) when I get this error.  In any event the fix on my end is easy for 
this item.

I ran into a second problem later on, here's the output from mdrun:

-------------------------------------------------------
Program mdrun, VERSION 3.3_beta_20041226
Source code file: ns.c, line: 314

Fatal error:
ns.c, 315: Negative number of short range atoms.
Call your Gromacs dealer for assistance.
-------------------------------------------------------

This was with the DPPC gmxbench system.  I get the same error with 
gmxbench's villin system.  These same input files work fine with our 
3.2.1 code.  Unfortunately my background is computer science not 
chemistry so this could be the result of a change in the behavior of 
grompp/mdrun that I need to learn about, but I didn't see anything in 
the manual (also checked-out from CVS) that referenced this.

The folks I work with are building a wrapper for GROMACS that automates 
running molecular systems on clusters.  There are some features in the 
current CVS cut that we would really like to take advantage of but we 
don't want to pester the developers with lots of questions if the 
codeline isn't in a state for this type of testing.  I got the sense 
reading the developers page at www.gromacs.org that it was pretty 
stable.  If you would rather we stick to 3.2.1 for now we can.

thanks,
charlie