The following gives instructions for fixing your build, which will
probably break the next time you update:
I'm currently working on a proposed implementation for binary trajectory
files in AMBER. The proposal uses the NetCDF library
(
http://www.unidata.ucar.edu/software/netcdf/) and I'm trying to
incorporate everything I've gathered from our discussions on this topic
over the past four AMBER developer meetings that I've attended.
I've just checked a limited, initial implementation into CVS. The next
time you update your sander directory, you will get a runmd.f and a
dynlib.f that reference code involved with netcdf and binary trajectories.
To fix your build, do a "cvs upd -d" from your $AMBERHOME/src directory.
This will get the new files you need, including the netcdf library
source. Next, you have to rerun the configure script. I haven't made any
changes to the config.h file, but it's necessary to run the netcdf
configure script prior to compiling -- the AMBER configure now calls the
netcdf configure with the proper arguments.
I have tested on ifort and g95. I have made efforts to make this
compatible with our other platforms as well, but so far this is
untested. I have reason to believe that further work may be necessary to
get this working with Absoft compilers.
I apologize for this disruption; I've tried to introduce this library as
cleanly as possible -- I hope this doesn't cause anyone too much
difficulty.
*********
IF YOU UNDERSTAND THE BUILD SYSTEM:
...then you're at least one step ahead of me. I think the way I've
hooked the configure scripts together is pretty clean, but I'm a little
lost with the sander Makefile. I've got it to a point where it seems to
work, but I'm not sure I've done that in the best way. I'm also not sure
how the makedepend script works. If someone who understands how this
makefile is supposed to work could check over my modifications, I would
be grateful.
Files changed in this checkin are
configure
Makefile
sander/Makefile
sander/runmd.f
sander/dynlib.f
*********
IF YOU'RE INTERESTED IN BINARY TRAJECTORY FORMATS:
I'm currently working with Dave, Tom (Cheatham), Carlos and Justin
Gullingsrud (of the VMD/NAMD camp) to design and implement an initial
framework for this. Once we have something reasonable, I plan to write
up and present a proposal to the larger AMBER community. If you feel
strongly that you'd like to be involved at these initial stages, please
contact me.
Thanks!
John
Received on Wed Apr 05 2006 - 23:49:51 PDT
