At 09:34 PM 10/30/98 EST, [log in to unmask] wrote:
>In a message dated 10/30/98 10:15:01 AM Pacific Standard Time,
>[log in to unmask] writes:
>
>> If the function is an extended intrinsic the main program should invoke the
>> intrinsic. If it isn't, then your routine should be invoked.
>
>I would expect many compilers to give my function priority over the
intrinsic;
>in fact, I thought the standard required this, so that introduction of new
>intrinsics would not break existing code.
Nope, actually it's the other way around. The introduction of new intrinsics
will potentially break old programs. That's one of the reason new intrinsics
tend to have long names; lessens the likelyhood of breaking something.
If you really want to be safe you should list all of your routines on
EXTERNAL statements and list all (vendor supplied extension0 intrinsics
on INTRINSIC statements. The latter will give you a compile time error
if you move to a compiler that doesn't support that particular intrinsic.
Hard to say whether or not that would be helpful. If one compiler supports
DTIME as an intrinsic and another merely has DTIME in some library
somewhere it's not likely to make any practical difference.
Dick Hendrickson
>That could be the subject of
>another test, to see which ways a compiler will permit replacement of
>intrinsic functions. I'll give this suggestion a try anyway.
>
>Thanks for a reply which is to the point.
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|