On Sat, Oct 31, 2015 at 3:25 AM, Mahmoud A. A. Ibrahim <
m.ibrahim.compchem.net> wrote:
> Dear All
> Thanks first for Marcelo and Hai for their support.
> Replacing OS is an impossible option in case of using high-performance
> computer.
Replacing the OS is never a suggested course of action (although Amber is
not really supported on Windows, any Linux variant should work fine).
> As well, installation of further python versions may result in
> deep technical issues related to other software.
>
This is not true. Python installations are isolated from each other.
There is no need to replace the system Python (version 2.6) in order to
install a Python version 2.7 or later, and it will not interfere with
anything using Python 2.6.
In fact, I've been convinced that no Python version less than 2.7 is worth
supporting anymore. Here is a nice post and explanation from a Red Hat
developer (who is also a core Python developer) explaining why Python 2.6
and older is not worth supporting:
http://www.curiousefficiency.org/posts/2015/04/stop-supporting-python26.html
With packages like Anaconda and Miniconda, installing a version of Python
2.7 or later has actually never been easier -- and you don't even need root
privileges to do it.
The problem of python2.6 with openmm package in AMBER14 would be solved by
> replacing the headline of reporters.py file with
> from __future__ import with_statement
> In our case, this solution works well and the test stage was passed
> successfully.
> So, we believe reporters.py should be updated in the source files of
> AMBER14.
>
Yes, this change will work for now. But ParmEd in the next version of
Amber will not work at all with Python 2.6. You will *need* to upgrade to
Python 2.7 or later. Most HPC clusters I've seen actually have a "module"
for Python 2.7. Something like "module load python/2.7" will drop a newer
version of python into your PATH.
All the best,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Oct 31 2015 - 09:00:04 PDT
