Re: Data Name = to more than one number
docdwarf_at_panix.com
Date: 02/10/04
- Next message: docdwarf_at_panix.com: "Re: Data Name = to more than one number"
- Previous message: Howard Brazee: "Re: Data Name = to more than one number"
- In reply to: JerryMouse: "Re: Data Name = to more than one number"
- Next in thread: Joe Zitzelberger: "Re: Data Name = to more than one number"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 10 Feb 2004 12:56:49 -0500
In article <t9GdnW0kV7Sbj7Td4p2dnA@giganews.com>,
JerryMouse <nospam@bisusa.com> wrote:
>docdwarf@panix.com wrote:
>> In article <IIydnQK5-PvJR7XdRVn-ug@giganews.com>,
>> JerryMouse <nospam@bisusa.com> wrote:
>>
>> [snip]
>>
>>> Doing different things based on the value of a single variable is the
>>> defining case for EVALUATE as in:
>>>
>>> EVALUATE ORDER-ITEM-SERIES
>>> WHEN '1' thru '2' (do something)
>>> WHEN '3' thru '4' (do something else)
>>> WHEN '5' CONTINUE (don't do anything)
>>> WHEN OTHER (do error processin)
>>> END-EVALUATE
>>>
>>> You really can't get more clear than that.
>>
>> With all due respect, of course, but 'clarity' is in the mind of the
>> beholder; it has been argued that 88-levels ('conditions') relay the
>> nature of processing to subsequent readers of the code in a manner far
>> superior to the hardcoding used above. Given the example above:
>>
>> 05 ORDER-ITEM-SERIES PIC X VALUE SPACES.
>> 88 VALID-SERIES-VALUE VALUES '1', '2', '3', '4', '5'.
>> 88 ITEMS-PRICED-NORMALLY VALUES '1', '2'.
>> 88 ITEMS-ON-SALE VALUES '3', '4'.
>> 88 ITEMS-DISCOUNTED-OTHER VALUE '5'.
>>
>> ...
>> EVALUATE TRUE
>> WHEN ITEMS-PRICED-NORMALLY
>> PERFORM NORMAL-PRICE-RITUAL THRU NPR-EX
>> WHEN ITEMS-ON-SALE
>> PERFORM SALE-PRICE-RITUAL THRU SPR-EX
>> WHEN ITEMS-DISCOUNTED-OTHER
>> PERFORM OTHER-DISCOUNT-RITUAL THRU ODR-EX
>> WHEN NOT VALID-SERIES-VALUE
>> PERFORM INVALID-ITEM-SERIES THRU IIS-EX
>> END-EVALUATE.
>>
>> (arguments about other aspects of the code given - upper case versus
>> mixed case, PERFORM THRU versus PERFORM, etc. - might best be left to
>> other threads)
>
>You make a good point. However, it must be noted that 88 levels are a
>half-assed attempt at logic and are to be avoided by right-thinkers.
As noted above - ''clarity' is in the mind of the beholder'.
Right-thinkers can, of course, use 88 levels sparingly...
... instead of liberally.
>I say
>"half-assed" because they only address half the logic problem.
Logic, as Wittgenstein tells us, is 'a game played by a particular set of
rules', not some manner of 'one size fits none' paradigm.
>
>For example:
>
>IF TIME-FOR-SUBTOTAL
> PERFORM SUB-TOTAL
> MOVE '0' TO WHITHER-SUBTOTAL <== WTF is "0"?
> END-IF
>
>Really should be:
>IF TIME-FOR-SUBTOTAL
> PERFORM SUB-TOTAL
> MOVE NOT-TIME-FOR-SUBTOTAL TO WHITHER-SUBTOTAL
> END-IF
This, it seems, is more of an argument against hardcoded literals than
against 88 levels; many's the program I've seen with
05 LIT-A PIC X VALUE 'A'.
05 LIT-B PIC X VALUE 'B'.
05 LIT-C PIC X VALUE 'C'.
...
05 LIT-9 PIC X VALUE '9'.
... used similarly.
>
>Even this tastes like a diet softdrink. Much clearer would be the construct:
>
>IF TIME-FOR-SUBTOTAL OF WHITHER-SUBTOTAL
> PERFORM SUB-TOTAL
> MOVE NOT-TIME-FOR-SUBTOTAL TO WHITHER-SUBTOTAL
> END-IF
See above about where clarity exists; a construct you neglect is
05 SUBTOTAL-FLAG PIC X VALUE 'N'.
88 TIME-FOR-SUBTOTAL VALUE 'Y'.
88 NOT-TIME-FOR-SUBTOTAL VALUE 'N'.
IF TIME-FOR-SUBTOTAL
PERFORM SUB-TOTAL
SET NOT-TIME-FOR-SUBTOTAL TO TRUE
END-IF.
(using SET... TO FALSE is an exercise left to the readers and
standards-writers; for the nonce I shall avoid an attempt to scale such...
... withering heights.)
>
>While your example above is quite clear to most, another complaint I have
>about 88 levels is they emphasise the business logic instead of the data
>logic.
That was, I thought, one of the strengths of COBOL... that a
carefully-crafted chunk o' code could be seen as clear by both a
programmer and ol' Louie in accounting.
>Consider your example rewritten with the following constants:
>
>01 FREE-VIBRATION-FLAG PIC X.
> 88 USE-FLEMING-CONSTANT VALUE '1'.
> 88 RANDOM-WALK VALUE '2'.
> 88 ENTROPY-ADJUSTMENT VALUE '3'.
> 88 PUNCTUATED-EQUILIBRIUM VALUE 'A'.
> 88 JANISSARY VALUE 'J'.
> 88 IGNEOUS VALUE 'F'.
> 88 MONROE-DOCTRINE-REDUX VALUE 'M'.
> 88 LEWINSKY-EFFECT VALUE 'S'.
>
>Who (besides "The Shadow") knows what's afoot?
Someone who is familiar with the language and needs of the business and
the process, a programmer who has spent a few years learning both code and
what the business she works for does...
... oh. Nobody is allowed to do that any more, that's right.
>
>
>This observation, however, comes to you from someone who doesn't use 88
>levels so should probably be ignored. Nothing here, move along.
[DIALECT HUMOR ALERT]
Suuuurrrree, an' whut's all this aboot loit'rrrin' rrround th' strrreets?
P'rrraps me good frrriend, Mr Bill E. Cloob, moight be teachin' ya
somethin' aboot th' meanin' of 'alacrrrity'?
DD
- Next message: docdwarf_at_panix.com: "Re: Data Name = to more than one number"
- Previous message: Howard Brazee: "Re: Data Name = to more than one number"
- In reply to: JerryMouse: "Re: Data Name = to more than one number"
- Next in thread: Joe Zitzelberger: "Re: Data Name = to more than one number"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
