Re Namelist IOSTAT behaviour

[Paul THOMAS]

IOSTAT should indeed yield a negative value because EOF has occurred with no error.  This is indeed what happens in gfortran, with the small program below.

If I change the &q to &p in the string written to unit 10, I get IOSTAT=-1 twice, followed by 9.

I think that we need to see the rest of your code before agreeing that there is a bug.

Paul T

$ cat test_eof.f90
program test_eof
  character*4 :: a
  namelist /q/ a
  open(10, status="scratch")
  write(10,*) "&q a=wxyz /"
  rewind(10)
  do j = 1,3
    a = ""
    read(10,nml=q,iostat=i)
    print *, a, i
  end do
end program test_eof

$ ./a
 wxyz           0
               -1
                9