Hi all,
There are a few test issues under cygwin. I am attaching files with
the relevant output, below will be a brief summary. There is one
problem with pbsa, a few with sleap, and a few for Rism. Here are the
tests which produce an error in the log:
cd pbsa_ligand && ./Run.t4bnz.min
Ouput shows:
NB-update: atom-based nb list 247422
SA Bomb in sa_arc()3: Allocates aborted 0 0
5014 0 0
cd ./rism1d/spc-mv0 && ./Run.spc
Output shows:
STOP 1
ERROR> RXRISM: reached steps limit Maxstep=10000
Here are the sleap tests which report diffs:
...
possible FAILURE: check mdl_1hpv.mol2.dif
/home/Dan/Amber/RC/amber11/AmberTools/test/sleap/pdbent
possible FAILURE: check hpv.prmtop.dif
/home/Dan/Amber/RC/amber11/AmberTools/test/sleap/amoeba
possible FAILURE: check leap.out.dif
/home/Dan/Amber/RC/amber11/AmberTools/test/sleap/molinfo
...
The first two tests actually do not run successfully (they segfault)
but don't generate errors in the log. They have failed in sleap under
cygwin since AT1.4. The last failure seems innocuous (a reported
torsion of 0 vs torsion of 360).
There are a bunch of Rism tests that report test failures, but don't
actually produce diffs, maybe because Rism uses the 'check1d' script
as a wrapper to dacdif? If you actually diff the files in question
there are in fact numerical differences, e.g.:
Dan.formula409:~/Amber/RC/amber11/AmberTools/test> diff
/home/Dan/Amber/RC/amber11/AmberTools/test/rism1d/spc-hnc/spc-lj.nvv
/home/Dan/Amber/RC/amber11/AmberTools/test/rism1d/spc-hnc/spc-lj.nvv.save
4003a4004,4005
> 9.99750000E+001 1.39381247E+005 1.39381247E+005 1.39380869E+005 1.39381247E+005 1.39381247E+005 1.39380869E+005 0.00000000E+000 0.00000000E+000 0.00000000E+000
> 1.00000000E+002 1.39485836E+005 1.39485836E+005 1.39485458E+005 1.39485836E+005 1.39485836E+005 1.39485458E+005 0.00000000E+000 0.00000000E+000 0.00000000E+000
The attached file contains a summary of all tests that failed. I'm
also attaching the raw 'log' and 'diff' files.
-Dan
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Wed Apr 13 2011 - 09:30:03 PDT
