Re: The joy of FORTRAN

On 3/4/25 3:42 AM, rbowman wrote:

On Tue, 4 Mar 2025 02:30:44 -0500, c186282 wrote:
 

On 3/4/25 1:25 AM, rbowman wrote:

On Mon, 3 Mar 2025 23:02:44 -0500, c186282 wrote:
>

     It's easy to kinda drift into that during development, you always
     think of ONE more thing you need to pass. However stage two, you
     put all that crap in a struct and pass one pointer. Each function
     can extract what it needs from that struct and ignore the rest.

>
I never made it to anywhere close to 60 parameters. We did have one
utility that took 22 command line parameters but it was never meant to
be invoked manually and came with a csh. Even then they all were
stuffed into a struct before the real function was called.
>
Even worse, the parameters were field widths in a positional record. I
truly hate positional records but the damn things live on.

>
    Hey, they have a function and certain simplicity.

 They are simple all right. It's always so much fun counting over to the
73rd character and knowing foo is contained in the next 5 characters.
Unless whoever constructed the record happened to leave a space out of the
value at the 52nd character that was supposed to be 8 characters and
everything past that is off by 1.
 But you save all those useless bytes for field delimiters!

   I pref delimiters - unless the individual fields are
   gonna be very small, thus meaning a large bulk is
   gonna be naught but delimiters.
   As for blanks ... the software writing the rec is supposed
   to check and PAD (or trunc) as necessary to ensure all goes
   in the right spots. Truncation SHOULD generate an error
   flag to warn the user - "Field 5, Alfred E. Neuman truncated
   to Alfred E. Neum" .... indeed the entry prompt should not
   provide room to enter anything longer than allocated for
   the field. Always assume user idiocy.
   Mostly we don't HAVE to use positional recs anymore since
   memory/storage has increased so much. Wasn't always that
   way however ....