Cobol - Re: Classic RW (was: Rounding errors

Author

Re: Classic RW (was: Rounding errors

Robert Wagner

2004-09-01, 3:55 am

On 30 Aug 2004 16:54:38 -0700, riplin@Azonic.co.nz (Richard) wrote:

>Robert Wagner <robert@wagner.net.yourmammaharvests> wrote

[color=darkred]
[color=darkred]
>But in any case, your two statements seemed to have the primary aim of
>denigrating 'the cognescenti' and 'academics'.

That wasn't my intention. Both encouraged the reader to learn what
others are saying.

>I was once a mainframe programmer when mainframes was just about all
>there were.

We all were. Some of us left the culture when it became practical to
do so.

>
>If your intent is to deliberately 'irritate people', then it may be
>working. But you are irritating the wrong people when they are
>constrained by their employers and terms of employment.

Hopefully, decision makers will p

in and learn how to save money,
make systems more reliable and programmers less cynical.

> Given the usual American foreign policy as I see it, then
>that may well be what they indoctrinated you with in the Marines.

I'm equally unhappy with American foreign policy. I'd like to see Iraq
turned into a modern country. Then citizens of other third world
countries would demand more from their governments. It's tragic that
billions of lives are wasted by inept or self-serving government.

>
>I think that it is very valuble to discuss such matters. I hope a lot
>of people are reading the messages and putting the results into the
>context of their own programming and what techniques they have been
>taught to use.
>
>However, if they think that 'Cobol has a bug' then the truth has lost
>out to 'tactical firepower'.

If they don't believe me, your barrage of flak will have prevailed.

>
>Which were valuable things to discuss. However you made claims that
>linked lists were much faster than tables as if this was a general
>statement of revealed truth when it was provavbly wrong.

It was another lesson learned.

>
>Specific case solutions are always faster than more generalised cases.

Good sort programs have more than one algorithm. I know SyncSort used
to have three or four.

>
>I have a different view on those issues. Reverting to 'what C did in
>the 80s' is not necessarily useful.

Professional programmers should know basics, like how to manage a
list.

>What you failed to present, probably because you never did much C, was
>why C++ was developed to solve the problems that existed in C, such as
>the problems inherent in using pointers. And then Java solved the
>problems inherent in C++.

Java syntax was copied from C++. Java solved the problem of
distributing software to multiple platforms.

>When you introduce the use of raw address pointers, and also of other
>things such as ENTRY, you completely fail to take into account the 20
>or 30 years of experience that C programmers have had with the
>problems that these introduce. Most likely because you weren't doing C
>in enough depth to care.

Namespace pollution isn't as bad as the mess created by mangling.
Cobol can have the same problems. I've worked on Cobol load modules
that had 100 callable programs.

>What would be more valuble is if you brought in these 30 year old
>techniques and explained why _not_ to use them. Then people would see
>why C++ was developed and would understand why OO is a valuble tool
>(but pointers are not).

I can't dispute that.

>I suppose that you will demand examples. Well just as a starter:
>"ANS84 does not define a standard file status code for 'file not
>found'".

In 74, it was a 9x. I forgot they standardized it in New Cobol.
Besides, I just write:
IF FILE-STATUS NOT EQUAL TO ZERO
DISPLAY 'Something is wrong ' FILE-STATUS
GO TO ABEND.

docdwarf@panix.com

2004-09-01, 3:55 am

In article <68o9j0hjmld6k3jurc9jps4or8onm6hpnq@4ax.com>,
Robert Wagner <robert@wagner.net.yourmammaharvests> wrote:
>On 30 Aug 2004 16:54:38 -0700, riplin@Azonic.co.nz (Richard) wrote:

[snip]

>
>Hopefully, decision makers will p

in and learn how to save money,
>make systems more reliable and programmers less cynical.

On the UseNet? Mr Wagner, you've asserted at other times that 'decision
makers' get all their ideas from articles in throw-away airline magazines
they have skimmed during cross-country flights... this kind of behavior
is, in my experience, at odd with the idea-generating stimuli reported by
UseNet readers.

What comes next... the plaintive wail of 'The children... errrrr, the
newbies! We must do it for the newbies!'?

[snip]

>
>If they don't believe me, your barrage of flak will have prevailed.

Hmmmm... so it seems that Napoleon was a bit neglectful; both God *and*
'truth' appear to be on the side of the biggest battalions.

[snip]

>IF FILE-STATUS NOT EQUAL TO ZERO
> DISPLAY 'Something is wrong ' FILE-STATUS
> GO TO ABEND.

Mr Wagner, this kind of code in an MVS-esque environment will cause ABENDs
where none need occur.

DD

Robert Wagner

2004-09-01, 3:55 pm

On 31 Aug 2004 22:05:08 -0400, docdwarf@panix.com wrote:

>In article <68o9j0hjmld6k3jurc9jps4or8onm6hpnq@4ax.com>,
>Robert Wagner <robert@wagner.net.yourmammaharvests> wrote:
>
>[snip]
>
>
>On the UseNet? Mr Wagner, you've asserted at other times that 'decision
>makers' get all their ideas from articles in throw-away airline magazines
>they have skimmed during cross-country flights... this kind of behavior
>is, in my experience, at odds with the idea-generating stimuli reported by
>UseNet readers.

Now that UseNet is indexed by Google, the probability of an
information s

er finding articles has increased.

>Hmmmm... so it seems that Napoleon was a bit neglectful; both God *and*
>'truth' appear to be on the side of the biggest battalions.

Napoleon would have beaten Wellington if the Prussians hadn't been
there.

The lesson is 'double-team Wagner'.

docdwarf@panix.com

2004-09-01, 3:55 pm

In article <90cbj09j17tso1adboevqhk8sp8m5uh882@4ax.com>,
Robert Wagner <robert@wagner.net.yourmammaharvests> wrote:
>On 31 Aug 2004 22:05:08 -0400, docdwarf@panix.com wrote:
>
>
>Now that UseNet is indexed by Google, the probability of an
>information s

er finding articles has increased.

Only if said 'information s

er' uses Google, Mr Wagner, and then only if
the Groups option is invoked... this kind of behavior is, in my
experience, at odds with the behavior usually exhibited by those who 'will
get all their ideas from articles in thros-away magazines they have
skimmed during cross-country flights'.

>
>
>Napoleon would have beaten Wellington if the Prussians hadn't been
>there.

My Sainted Paternal Grandmother - may she sleep with the angels! - would
have been a trolley-car if she had had wheels.

>
>The lesson is 'double-team Wagner'.

The lesson might be ''truth' is *not* on the side of the greatest
battalions'... or, as Galileo is reputed to have said, 'Eppur si muove'.

DD

Howard Brazee

2004-09-01, 8:55 pm

On 31-Aug-2004, docdwarf@panix.com wrote:

>
> Mr Wagner, this kind of code in an MVS-esque environment will cause ABENDs
> where none need occur.

Certainly. We had a program which we were supposed to use when doing IO
commands with a KSDS file. That program returned a "7" when it found a file
status not equal to zero. The calling program didn't know what to do with that
7 so it normally abended. The abend didn't show the real error.

I found out that often times, the *real* error was "97", which means "For VSAM
only: OPEN statement execution successful: File integrity verified".

Now our shop doesn't work by standards of OO shops. When we make a change all
programs effected need to be extensively unit tested, systems tested, and user
tested before implementing. Because of this, we never fixed the program to
accept "97" as valid.

The sin the original programmer did wasn't overlooking "97". His sin was in
hiding the return code, leaving the calling program in the dark.

Sponsored Links