On Mon, Feb 6, 2017 at 10:30 AM, David Case <david.case.rutgers.edu> wrote:
> On Mon, Feb 06, 2017, Ross Walker wrote:
>
> > So does this mean we are officially dropping RHEL 6 support?
>
> No. Amber users are provided (via conda) with the correct python
> (currently version 2.7). Amber codes will not use the system-supplied
> python,
> and hence won't care what version (if any) RHEL supplies.
>
> Note: there are still some glitches involved in this transition, but I
> think
> we are close.... Note that almost all users have been downloading their
> python from Contiuum since April of 2016, and so have not been relying
> on the system python for some time.
>
> [For personal reasons, some Amber users/developers may wish to
> use another python version. Since they are experts in python, they are
> welcome to do this. Much of the recent back-and-forth about python
> involves
> how much hand-holding we should provide for people who don't want to use
> the python we supply.]
>
This is a slight wrinkle, I think. Unlike some other packages (like
OpenMM), Python is not central to Amber's functionality, and there are
realistic use-cases to not worrying about Python at all. If someone uses
--skip-python, I think it's highly desirable that the build not then fail
with a Python-related issue :).
There are *some* components that depend on Python in all cases. A notable
example is update_amber, which I wrote to be Python 2.4 -- Python 3.x
compatible specifically so it would run anywhere. Since what Hai wrote was
a small utility to do some post-processing, that should be done in a way
that doesn't depend on having a specific version of Python (or any
dependencies) if at all possible, in my opinion.
Thanks,
Jason
--
Jason M. Swails
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Mon Feb 06 2017 - 11:00:03 PST
