Hi,
iargc() should be replaced with command_argument_count().
The wiki aludes to these problems, but does not yet indicate the
best current practice which is described below from my cvs log.
http://ambermd.org/pmwiki/index.php/Main/Coding
Date: Sat May 23 19:36:48 2009 +0000
Replaced gnu iargc() with f2003 standard command_argument_count().
The same could be done for getarg. Some compilers like gnu already
support these f2003 routines in preference to the old gnu ones:
http://www.nag.co.uk/nagware/np/r51_doc/relnotes.html
http://docs.sun.com/source/819-3684/2_F95_Intrins.html
However, some compilers directly support neither:
pgf90 7.0-4 64-bit target on x86-64 Linux
but later versions only support the f2003 routines:
pgf90 8.0-5 64-bit target on x86-64 Linux -tp k8-64e
cpp -traditional -P -DBINTRAJ -DSQM sqm.f > _sqm.f
pgf90 -c -O1 -Mfree -o sqm.o _sqm.f
PGF90-S-0038-Symbol, iargc, has not been explicitly declared (_sqm.f)
0 inform, 0 warnings, 1 severes, 0 fatal for sqm
Scott
On Wed, Oct 06, 2010 at 11:48:38AM -0300, Gustavo Seabra wrote:
> Works now. Looks like PGI is being even stricter than gnu...
>
> On Wed, Oct 6, 2010 at 11:18 AM, Jason Swails <jason.swails.gmail.com> wrote:
> > i just added iargc to the narg integer declaration line -- try now? It
> > wasn't giving problems on gfortran or ifort
> >
> > On Wed, Oct 6, 2010 at 10:13 AM, Jason Swails <jason.swails.gmail.com>wrote:
> >
> >> I may have... I did try compiling on kraken with the PGI compilers. Let me
> >> check (my fault, since it's my source file anyway)
> >>
> >> On Wed, Oct 6, 2010 at 9:59 AM, Gustavo Seabra <gustavo.seabra.gmail.com>wrote:
> >>>
> >>> I'm trying to compile AmberTools serial using PGI compilers, from a
> >>> freshly cloned branch (pulled from master a few minutes ago), but
> >>> compilation is dying with this error:
> >>>
> >>> PGF90-S-0038-Symbol, iargc, has not been explicitly declared (calcpka.f90:
> >>> 114)
> >>> 0 inform, 0 warnings, 1 severes, 0 fatal for calcpka
> >>> PGF90-S-0026-Unmatched quote (calcpka.f90: 331)
> >>> 0 inform, 0 warnings, 1 severes, 0 fatal for usage
> >>> make[1]: *** [calcpka] Error 2
> >>> make[1]: Leaving directory `amber/AmberTools/src/etc'
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Oct 07 2010 - 12:30:03 PDT
