Re: DOW vs DOU- Ha

RE:     Re: DOW vs DOU- Ha

On May 28 1997 Jim Welsh wrote:

>>I prefer this method John.
>>
<<<  Blah blah snip  >>>> 
>>  KEY      CHAIN  FILEXXX
>>  *IN90    DOWEQ  *OFF                        While a record
>>  XXXCODE  ANDNE  'B'                         & not code B
>>           EXSR   CHKRCD                      Check Record
>>  GOODR    IFEQ   'Y'                         Process ?
>> <.... whatever process ....>
>>           ENDIF                              End Process
>>  KEY      READE   FILEXXX                 90 Read Next
>>           ENDDO
>>
>>C* Check Record Subroutine
>>  CHKRCD   BEGSR
>>           MOVE 'Y'        GOODR
>>  XXXCODE  IFEQ 'A'
>>           MOVE 'N'        GOODR
>>           ENDIF
>>           ENDSR

>This works fine when there are only 2 data types, A and not A. The problem 
>with this approach, however, is that type A does not really equate to "not 
>good", and type not A does not equate to "good". Each is a "good" type 
>within the iteration of mixed type A and type not A records. Type A records 
>simply have no processing to be done.

>Later on the spec. might change and we would have to perform some work on 
>type A records after all. Or we may need to do different processing for type 
>X and type Y. That's why SELEC or CAS are the best codes to use within the 
>loop. They are an exact match for the structure of the data. (I've gone 
>right off ITER in the last couple of days for obvious reasons.) :-)

<snip>
>Dave Kahn - Tengizchevroil, Kazakstan
--------------------------------------------------------------------------
Dave;
That is correct. 
  
The "A" type records are valid selections, They just don't have any further 
processing to be done(TODAY).  The example's intent (HA, I still have 
to laugh, I'm sorry. But you'd have to admit it was funny. BTW I told
charlie(who also made the mistake I believe) that I had to print yours off 
and read it over acouple of times with a cup of coffee just to enjoy it.
Yes I intentionally "put the pail over the door" , thanks for opening it), 
aaaanyway where was I, the example's intent was just for your 
benefit(so to speak).

The rational that I believe is most important is to 
"Separate the Controling Logic from the Data Manipulation Logic!"  

If I really had a senerio where, while(until imean:-)) reading a file
I had some records to process, and some not to. I'd perform a subroutine
for the records to be processed.  I would not mix (with few exceptions) 
data manipulation(adds,divides, Eval's, etc.) with the controlling logic
(If, else, when, case, etc.). This makes enhancements much easier, New
selects or other conditions could be added easily in their section, and
other data manipulations could be done in their section.

"Think of the s/he comming behind you"  Enhanceability is #1

P.S.  I orginally used "FILEXXX" and XXXCODE"  so that
people would see that the variable came from the record just read.
In some cases I guess I was wrong.

John Carr CDP
EdgeTech
5921 Acorn Ridge CT.  Midlothian VA 23112
74711.77@compuserve.com 
Ph 804-739-7689  Fax 804-739-7950

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* This is the Midrange System Mailing List!  To submit a new message,   *
* send your mail to "MIDRANGE-L@midrange.com".  To unsubscribe from     *
* this list send email to MAJORDOMO@midrange.com and specify            *
* 'unsubscribe MIDRANGE-L' in the body of your message.  Questions      *
* should be directed to the list owner / operator: david@midrange.com   *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *