topic Thank you for your answers. I in Intel® Fortran Compiler

loop question

eide — Tue, 16 Oct 2012 08:31:30 GMT

Hi,

please see the following test program.

[fortran]

     program counter
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!
!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
   implicit none
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
   integer (kind = 4)     :: icel,ncc

! initiate
   ncc = 10
! loop
   do icel = 1,ncc

       write(*,*)icel,ncc

   enddo ! icel
!
   write(*,*)'================================='
   write(*,*)icel,ncc

   end program counter

[/fortran]

[plain]

screen output

           1          10
           2          10
           3          10
           4          10
           5          10
           6          10
           7          10
           8          10
           9          10
          10          10
=================================
          11          10

[/plain]

Why is icel one more after the loop? I am not new to fortran, but I have never observed that before. I am using using composer_xe_2011_sp1.9.293 on an 64bit opensuse 12.1 with kernel 3.1.10-1.16. If someone could explain that to me, would be fine!

Eide

Why is icel one more after

mecej4 — Tue, 16 Oct 2012 10:06:59 GMT

Why is icel one more after the loop?
Because that is what the Fortran Standard specifies. See the explanation at http://software.intel.com/sites/products/documentation/doclib/stdxe/2013/composerxe/compiler/fortran-win/index.htm or see section 8.1.6.6.2-1(3) of the Fortran-2008 Standard.

This behavior was defined in

TimP — Tue, 16 Oct 2012 12:19:09 GMT

This behavior was defined in the f77 standard. Until 30 years ago, a compiler I used produced 3 different results for the value of the loop counter after loop termination, depending on optimization level.

Thank you for your answers. I

eide — Tue, 16 Oct 2012 18:54:42 GMT

Thank you for your answers. I was not aware of that before. I stumbled on this while running into a segmentation fault.