RE: RPG Line Numbers: was externally defined D/S

John,

The line number in the compiler listing should match the source member. In 
fact, so should the debug code.

If you're (generally speaking, of course, not "you" specifically John) dumb 
enough to optimize your code AND include debug in it, you get what you pay 
for.  The RPG messages at the bottom of compiler listing usually have the 
correct statement number. The compiler should just put those same number to 
the left of the stmt, and the debugger should work with them. The external 
stuff should have some kind of prefix like we had with the "real" rpg 
compiler on the System/38.

Even with optimization, there is nothing keeping the developers from 
reporting a runtime error with the source statement number. Even if that 
number is consolidated, there is a pretty good "guess" that can be made to 
determine the correct number. Heck, our good friend Ken Kelly even 
dis-assembled RPGIII from the program template... restoring the _original_ 
source lines numbers. If Ken could do that, how hard can it be to put the 
right numbers in RPG IV objects to begin with.

I think the compiler listing was ported and put into production with much 
enhancement. It should be fixed, but to tell you the truth, I'd rather have 
them get rid of the COLON separator and add FILE HANDLES to the language 
first.

Bob Cozzi



On Thursday, June 26, 1997 8:35 PM, John Carr 
[SMTP:74711.77@CompuServe.COM] wrote:
>
>
> Date: 6/26/97  8:56 PM
>
> RE:   Re: Sources for External Data Structures
>
> >I am responding to Hans Boldt comments on ILE.  I think the basic 
concept of ILE
> >is great.  However, I think it is ugaly.  Ugaly is 1,000 times worse 
than ugly.
> >As a developer, it is hopeless.  If a client calls up with a problem and 
u have
> >to debug the pgm, good luck. What line number do u use?   Do you want to 
give
> >all your clients your source, NO.  We worked with a client and they keep 
saying
> >they had a problem.  We had NO way of debuging.  We compiled the pgm and 
shipped
> >them the new pgm and SRC.  Now we went into debug mode after 3 hours and 
found
> >out that it was a user error.  UGH.  If this is progress, then we need 
the
> >rosetta stone back.    ILE is OK for inhouse operations.  It is very 
very very
> >bad for software vendors.  IBM should look at debugging as a major 
problem and
> >get their head out of the sand.  Line number are very, very very, very, 
very
> >very important.  However, Toronto must not think so.
> >Peter H. O'Connor
> >PAE
>
> Peter
> I share your frustration.  I remember the first time I saw what they were
> doing and didn't like it either.
>
> However the design team did have a serious dilemma.  Lets suppose you 
have
> a line of calc's that says;
>
> ADD   1   X
>
> And you do not reference X anywhere else in your program.  If you 
optimize
> the program, the compiler may throw out that line of calcs( there are
> many similar situations involving intermediate result fields, etc)
>
> Now you start Debug and put a break point on a line of calcs that DOES
> NOT even exist in the Object code!  What would you like it to do?
>
> The compiler might even "move lines of calcs around" during the optimize
> process!  "What you see may not be what you have"



Bob Cozzi
Bob@rpgdev.net
http://www.rpgdev.net


* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* 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   *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *