On Oct 16, 2010, at 10:42 AM, case <case.biomaps.rutgers.edu> wrote:
> Jason:
>
> What was the final result of thinking about not requiring python 2.6 going
> forward? My group is reporting lots of trouble installing that version, and
> my vague recollection was that there might be a way to avoid it(?). Can you
> summarize the current requirements for the master branch in git?
I have removed the python 2.6 requirement in the git tree, and I've tested that it works with versions as old as 2.4, which is as old a version as any of the Linux distros I've tried it on has. It should pass all of the tests just fine now with any Python version 2.x.
>
> Sorry for the vague problem report; I'm just trying to get my head around
> where things stand with this. Do we need an option, like the -noX11 option to
> configure, that will allow users to build Amber without mmpbsa_py? This might
> be useful on supercomputers or test machines, where we aren't going to use
> mmpbsa_py anyway, but can't easily build the rest of the codes unless the
> correct python is available.
I can certainly add this in, but I don't think it's necessary. The MMPBSA.py install is done in such a way that failures are non-fatal, which should be good enough, although having the build log littered with make failures may be unsightly.
There are still advantages to including our own python version, though. Python 3 is a significant diversion from 2.x. Whereas 2.4 code is compatible with 2.6, 2.6 code is largely incompatible with any Python 3.x version. We can also hide useful python modules away in the lib directory so they appear as system modules, and it will alleviate the problems in getting things working while cleaning up "junk" in the AMBERHOME/bin directory. Ben and I are also working on creating a library of useful python utilities for amber which could be included, and installing our own python can help us keep track of dependencies and make the install process more painless.
A final question or so: what is your group having problems building? Python or the git version of MMPBSA.py? If it's the Python version, did they try the configure_python script I added alongside the configure_openmpi script?
Thanks!
Jason
>
> ...thanks...dac
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
--
Jason Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Sat Oct 16 2010 - 08:30:03 PDT
