Re: amber-developers: bench.jac fails in for amber9

From: Scott Brozell <sbrozell.scripps.edu>
Date: Thu, 9 Mar 2006 12:38:36 -0700

Hi,

Bob is right.
The plan for retired namelist variables was to leave them in the namelist
and emit warnings for inputs that still define them.
This is the only way to avoid errors like the below on all platforms.
The Retired Namelist Variables appendix lists all such variables.

In creating the list I noticed that most of the sources did not follow
the plan. So I shall patch the sources; I have started on sander's
cntrl namelist, but I won't get this done until early next week.

For those interested in the right way:
1. leave the retired variable in the namelist, but move it to the tail
   with all the other retired variables.
2. assign the variable the default value RETIRED_INPUT_OPTION.
3. after the namelist is read, check the variables value and issue
   a warning if it has changed.

For example, from sander's mdread.f

...
   _REAL_ timlim ! retired
...
   namelist /cntrl/ irest,ibelly, &
...
         dtemp, dxm, heat, npscal, timlim ! all retired
...
   timlim = RETIRED_INPUT_OPTION
...
   if ( timlim /= RETIRED_INPUT_OPTION ) then
      write(6,'(/,a,/,a,/,a)') 'Warning: timlim has been retired.', &
            ' Check the Retired Namelist Variables Appendix in the
manual.'
   end if

Scott

On Thu, 9 Mar 2006, Robert Duke wrote:

> I believe that a,b,c, and alpha beta gamma were intentionally pulled
from
> the &ewald input in sander 9; I don't see them documented and remember
> reading a comment about pulling them because they were dangerous in the
> code. My guess is that whoever made this decision forgot to update
> everything in sight in the test and benchmark dirs. I actually have not
> pulled this stuff from pmemd; I am more worried about other things.
> Regards - Bob
>
> ----- Original Message -----
> From: "Andreas Svrcek-Seiler" <svrci.tbi.univie.ac.at>
> To: <amber-developers.scripps.edu>
> Sent: Thursday, March 09, 2006 2:08 PM
> Subject: amber-developers: bench.jac fails in for amber9
>
>
> > Dear developers,
> > The subject says nearly all.
> > trying bench.jac, I got
> > lib-4324 : UNRECOVERABLE library error
> > The variable name 'A' is unrecognized in namelist input.
> > ...from pathscale-compiled sander
> > and
> > invalid reference to variable in NAMELIST input, unit 5, file
> > /scr/orgon/svrci/WORK/DEBUG/amber9/benchmarks/jac/mdin, line 12,
position
> > 3
> > ...from the ifort-compiled version.
> >
> > So it's probably not a compiler error.
> > The line pointed to above
> > reads
> > a=62.23, b=62.23, c=62.23,
> > so "a=..." night be the offending expression
> >
> > best regards,
> > Andreas
> >
> >
> >
> >
> > --
> >
> > )))))
> > (((((
> > ( O O )
> >
-------oOOO--(_)--OOOo----------------------------------------------------
-
> > o Wolfgang Andreas Svrcek-Seiler
> > o (godzilla)
> > svrci.tbi.univie.ac.at
> > .oooO Tel.:01-4277-52733
> > ( ) Oooo. -------\
> > ---( )--------------------------------------------------------
> > \_) ) /
> > (_/
> >
> >
>
>
Received on Wed Apr 05 2006 - 23:49:41 PDT
Custom Search