I use -nobintraj because the build & config goes faster and (normally)
is more likely to work. Perhaps today marks the tipping point where
the reverse has become true.
Josh
On 13/02/2014, Jason Swails <jason.swails.gmail.com> wrote:
> On Thu, Feb 13, 2014 at 9:42 AM, David A Case
> <case.biomaps.rutgers.edu>wrote:
>
>> On Thu, Feb 13, 2014, Josh Berryman wrote:
>>
>> > Hello, I just built the head with these flags:
>> >
>> > ./configure -noX11 -nobintraj gnu
>> > make -j 8 install
>> >
>> > and the mdgx build failed at link time with:
>> >
>> > /usr/bin/ld: cannot find -lnetcdf
>> >
>> > ...which is not surprising given that "-lnetcdf" (and -lfftw3) are
>> > hard coded in the mdgx Makefile.
>> >
>> > I've just switched this to $(NETCDFLIB) and $(FLIBS_FFTW3) . Anyone
>> > who knows more about (mdgx/AmberTools/life) should change it back
>> > again quick if they think that I have only made things worse here.
>>
>> No...this won't work. The libmdgx.so target needs to have a relocatable
>> netcdf (so that it can be linked to python code), but $(NETCDFLIB)
>> generally
>> points to libnetcdf.a.
>>
>> My vote would be to remove the -nobintraj option. We've been using
>> bintraj
>> for trajectories for years and now use it for restarts as well. I think
>> features of REMD require it. So: why do you want to use that option?
>>
>> It's possible to add a lot of extra logic to configure2, and yet more
>> defines in config.h. But you can see the problems that arise when there
>> are flags (like -nobintraj) that are almost never used. Do you have no
>> workaround?
>>
>
> Ick, more configure2 logic. :) The workaround is to specify -nofftw3 to
> avoid building mdgx (and RISM) in the first place. I've reversed the mdgx
> Makefile change and tied the -nobintraj option to disable the mdgx build
> for now to avoid this problem.
>
> I agree with Dave that unless a good use-case can be found for disabling
> the best way we have to store trajectories (and restarts), -nobintraj
> should be deep-sixed.
>
> All the best,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Feb 14 2014 - 00:00:03 PST
