The ABINIT Users Mailing List ( CLOSED )

[Alain Jacques <ajs.jacques@gmail.com>]

Fell on this one while compiling abinit-5.6.0 on OSX x86 10.5.4 with
gcc-4.3.2:

Making all in 15recursion
/opt/openmpi-1.2.6_gcc-4.3.2/bin/mpif90 -DHAVE_CONFIG_H -I.
-I../../../src/15recursion -I../.. -I../../src/defs -I../../../src/defs
-DNDEBUG -DNDEBUG   -O3 -march=core2 -c -o recursion.o
../../../src/15recursion/recursion.F90
../../../src/15recursion/recursion.F90:410.23:

&     mult*erreur(irec)
                      1
Error: Nonnegative width required in format string at (1)


and the corresponding code is

!structured debugging if prtvol=-level : print detailled result the
first time we calculate density/kinetic energy
!if(prtvol==-level)then
 if(get_rec_coef/=0)then
  if(first(get_rec_coef)==0)then
   write(message,'(a,3i3)') ' coordonnees',coordx,coordy,coordz
   call wrtout(06,message,'PERS')
   write(message,'(a)')'  irec, densite, erreur_theorique'
   call wrtout(06,message,'PERS')
   do irec=0,nrec
    if(bn2(irec)>tol14)then
     write(message,'(i,2d10.3)') irec,  0.5d0*mult*real(acc_rho(irec),dp), &
&     mult*erreur(irec)
     call wrtout(06,message,'PERS')
    else
     exit
    end if
   end do
   first(get_rec_coef) = 1
!  DEBUG
!  write(6,*)'a', an
!  write(6,*)'b2', bn2
!  ENDDEBUG
  end if
 end if
!endif

My gfortran doesn't seem to like empty integer field width i.e.
     write(message,'(i5,2d10.3)') irec, 
0.5d0*mult*real(acc_rho(irec),dp), &
compiles nicely.
Anyone with a suggestion for a relevant field width here? Too late to
think by myself.

Kind regards,

Alain