Re: [AMBER] AmberTools12 unset variable in sqm/MNDOChargeSeparation.F90

From: Andreas Goetz <agoetz.sdsc.edu>
Date: Mon, 22 Oct 2012 15:02:28 -0700

On Oct 22, 2012, at 1:18 PM, Ake Sandgren wrote:

> On Mon, 2012-10-22 at 13:11 -0700, Andreas Goetz wrote:
>> Hi Ake,
>>
>> You are right, the variable dipfac is never set. However, none of the other variables that depend on it are used. Thus, while the code is not clean, it produces correct results.
>>
>> What is the problem you see with the test suite?
>
> No problems as such. It's just that the intel compiler complained about
> it when i ran it with -check=uninit.
>
> I'm producing a bunch of patches to fix uninitialized usage and other
> compiler warnings.
>
> Interested?

Very much :-) You can email me patches offlist so I can check/integrate these with the development tree.

I have noticed that the intel compiler can issue false positive warnings, in particular if common blocks are involved. Please be aware that this might require going through the code manually.

Thanks,
Andy

PS: Speaking of common blocks, we would love to see those replaced with modules, in case you have any spare time ;-)


>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber

--
Dr. Andreas W. Goetz
Assistant Project Scientist
San Diego Supercomputer Center
Tel  : +1-858-822-4771
Email: agoetz.sdsc.edu
Web  : www.awgoetz.de
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Oct 22 2012 - 15:30:03 PDT
Custom Search