FIX: "Integer divide by 0" Error Handling

Article translations Article translations
Article ID: 118337 - View products that this article applies to.
This article was previously published under Q118337
Expand all | Collapse all

On This Page


In code, there are two significant problems with dividing by 0:

  • "Integer divide by 0" is not generated as a run-time error. Instead, one of two error messages is produced:

    • Under MS-DOS, the DOSXMSF error message is produced:
      DX1020 unhandled exception: divide by zero exception
    • Under Windows NT, the Dr. Watson error message is produced:
      exception: divide by zero
  • With optimization on, the compiler replaces any constant expression that contains an "Integer divide by 0" with 0, which is incorrect, and produces the following error message:
    warning F4723: potential divide by 0
    (With optimization on, but warnings off, the compiler fails to generate an error or warning, either during compilation or run time.)


Microsoft has confirmed this to be a bug in the Microsoft products listed at the beginning of this article. This problem was corrected in Microsoft FORTRAN PowerStation 32, version 4.0.


To generate the "Integer divide by 0" exceptions, compile and run the "Sample Code 1," below. With optimization off, the exception occurs on the first print statement. With optimization on, it occurs on the second print statement.

NOTE: FORTRAN PowerStation for MS-DOS, version 1.0, correctly generates error F6003 in this case.

To produce the invalid code generation and warning F4723, compile and run either sample with optimization on (option /Ox). To demonstrate invalid code generation with no warnings or errors, compile and run "Sample Code 2," below, with optimization on and warnings off (/Ox /W0). With optimization off (default option), "Sample Code 2" will correctly generate the run-time error for "Integer divide by 0": M6103.

NOTE: /4Yb ($DEBUG) does not produce any additional error messages.

Sample Code 1

c compile options needed: See text above.

      integer i,j
      character*4 c /"0"/ 
      print *, ' 1/0 =',j/i
      read(c,'(I1)') i
      print *, ' 1/0 =',j/i

Sample Code 2

c compile options needed: See text above.
      real a,b
      print *, ' 1.0/0.0 =',b/a


Article ID: 118337 - Last Review: October 24, 2003 - Revision: 2.0
  • Microsoft FORTRAN PowerStation 1.0 Standard Edition
  • Microsoft Fortran PowerStation 1.0a for MS-DOS
  • Microsoft FORTRAN PowerStation 32
kbbug kbcompiler kberrmsg kbfix KB118337
Retired KB Content Disclaimer
This article was written about products for which Microsoft no longer offers support. Therefore, this article is offered "as is" and will no longer be updated.

Give Feedback


Contact us for more help

Contact us for more help
Connect with Answer Desk for expert help.
Get more support from