On Wed, Nov 09, 2016, Neale, Christopher Andrew wrote:
> As reported in this post:
> http://dev-archive.ambermd.org/201609/0004.html . pmemd fails for a
> number of steps larger than MAXINT (see below). I see the logic in
> Ross' reply, but I think it would be useful if amber detected this and
> gracefully exited with a notice that nstlim has to be less than MAXINT.
>
> $ grep nstlim md.in
> nstlim=2500000000,
>
> $ pmemd -O -i md.in -o MD1.out -p this.prmtop -c EM.rst -r MD1.rst -x MD1.mdcrd -inf MD1.info -l MD1.log
> At line 639 of file mdin_ctrl_dat.F90 (unit = 5, file = 'md.in')
> Fortran runtime error: Integer overflow while reading item 0
>
This would be a lot harder than you (might) think to code. The reading of
namelists is done by code provided by the compiler writers, not by amber
developers. Different compilers vary a lot in the user-friendliness of their
error messages.
There are various approaches that could be taken to get a better error
message, but the message above is not all that bad, in my view: It points you
to the exact line in the code where the error occurs, and indicates that
it is an integer overflow. There are *lots* of other places where Amber
fails to make good checks for bad inputs, and fixing those should probably
have a higher priority.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 09 2016 - 15:00:02 PST
