On Tue, Aug 18, 2015 at 3:50 PM, Adrian Roitberg <roitberg.ufl.edu> wrote:
> I agree.
>
> But then the test cases will have a fixed ig ?
Yes.
> I am all for it, but....
>
> I am very afraid of people lifting the mdins from the test cases and
> forgetting to change ig...
>
Still better than what we have now ;). None of the test cases use ig=-1
now (whether they use the default or specify an explicit seed) -- nor can
they. So comparing before and after this change:
before - people lift mdin files that may or may not specify ig=# (but
never -1 for tests). They don't change it, get the same seed every time...
bad things
after - people lift mdin files that *always* have ig=# (but never -1 for
tests). They don't change it, get the same seed every time... bad things.
So I fail to see how changing ig=-1 makes *this particular* concern any
worse than it currently is. But it makes the default overall much better.
Unfortunately, it's impossible to alleviate this concern, since we *must*
use ig>0 for the tests, since they're regression tests.
> I guess it is a case of foolproofing creating more fools ?
>
Could be...
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 Tue Aug 18 2015 - 13:30:05 PDT
