It seems innocuous enough to ask that the standard define names for
values of the IOSTAT variable for errors mandated by the standard.
However, one compiler that I used packed together the error code,
the unit number, and the type of I/O statement into one integer.
Mandating a named constant for the error code would preclude such
shenanigans, which may be a good thing or a bad thing.
If the standard specifies that the error is identified by a named
constant, but not that the value of the IOSTAT variable will be the
same as the named constant. such shenanigans would still be permitted.
I.e., one could do mod(my_status,512) and compare it to the standard-
mandated list of error codes.
Best regards,
Van Snyder
|