The ABINIT Users Mailing List ( CLOSED )

[Paul Fons <paul-fons@aist.go.jp>]

Wanting to see how far I can go in compiling abinit today, I  
commented out the (two) lines in rhorferm3.F90 (that related to  
debugging the parallel version) and started make again with fingers  
crossed.  Looks like the same problem exists in the routine ./ 
Src6_response/vtorho3.F90.  There is an INTENT statement in both  
routines that relates to an outgoing argument in which the second  
argument is a constant (rather than an integer variable).  Is this  
just silliness on the part of the compiler (and would a sufficient  
hack be to define a local integer variable in the lines write before  
the write statements ?)




/usr/bin/cpp  -Dmacosx -DFLOAT="double precision" -DPOINTER_SIZE=4 -P  
vtorho3.F90 > vtorho3_cpp.f90
f90  -w -O -f free -N11 -cpu:g5 -altiVec  -I../Lib_nanoquantaxc -c  
vtorho3_cpp.f90

    write(message,'(i5,a,2es13.6)')ifft,' ',vtrial1(ifft,1),rhor1 
(ifft,1)
                                                                        
 ^
cf90-1640 f90fe: ERROR VTORHO3, File = vtorho3_cpp.f90, Line = 272,  
Column = 73
  Dummy arguments with the INTENT(OUT) attribute must be defined  
before use.

     write(message,'(a,2es13.6)')'      ',vtrial1(ifft,2),rhor1(ifft,2)
                                                                      ^
cf90-1640 f90fe: ERROR VTORHO3, File = vtorho3_cpp.f90, Line = 275,  
Column = 71
  Dummy arguments with the INTENT(OUT) attribute must be defined  
before use.