read float into integer

[pkienzle <pkienzle at gmail dot com>]

I don't know if it is a bug, but gfortran does not truncate floating point
values toward 0 on input.  The f90 code I'm trying to compile assumes that
it does.

Here's a small test program to demonstrate the issue:

      integer(4) :: n
      logical    :: b
      read (*,*,err=20) n,b
      write (*,*) n,b
      stop
 20   write (*,*) 'Couldn''t read integer.'
      end

Here is the comparison between Intel Fortran and gfortran on OS X:

$ifort io.f90 
$echo 2 T | ./a.out
           2 T
$echo 2.9 T | ./a.out
           2 T
$echo 2.900e3 T | ./a.out
        2900 T
$echo -2.9 T | ./a.out
          -2 T
$gfortran io.f90 
$echo 2 T | ./a.out
           2 T
$echo 2.9 T | ./a.out
 Couldn't read integer.
$echo 2.900e3 T | ./a.out
 Couldn't read integer.
$echo -2.9 T | ./a.out
 Couldn't read integer.


-- 
View this message in context: http://www.nabble.com/read-float-into-integer-tf4350081.html#a12394785
Sent from the gcc - fortran mailing list archive at Nabble.com.