On Wed, Mar 11, 2015 at 7:57 AM, Brent Krueger <kruegerb.hope.edu> wrote:
> I'll just add another voice to represent those who work with older
> systems. If nothing else, I'll confirm that my group is one who actively
> uses a 4.1.2 system all the time as our primary system and there are no
> plans to upgrade this anytime soon.
>
> I think that most AMBER developers are much closer to the edge with regard
> to O/S upgrades than the AMBER user community, so what is old for most
> developers is 'stable' to most of the users. Maintaining support for O/S
> that are 5-ish years old (e.g. CentOS 5.6) is not crazy in my opinion.
>
But you can install gcc44 in CentOS 5.6 using a simple "yum install
gcc44", right? It's currently a little inconvenient to specify this
compiler instead of the 4.1.2, since the compilers are named gcc44, g++44,
and gfortran44. But if we had configure2 look for and use those compilers
instead of the 'standard' ones if GCC 4.1.2 was detected, would that
satisfy your needs?
For the time being, gcc 4.1.2 is supported (and based on feedback I don't
plan on pushing to remove that support for AmberTools 15), but I use the
word 'supported' in a loose manner. RISM doesn't work. The file-less
sander API doesn't work. EMAP restraints don't work. mdgx and the pbsa
FFT solver don't work. The majority of the rest of the tests pass, but it
will be difficult to determine what is going to fail and have good
mechanisms in place to fail gracefully and informatively.
So while it's "supported", I certainly can't recommend it.
Thanks for the feedback,
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
Received on Wed Mar 11 2015 - 05:30:02 PDT
