Re: pointer problem

[Scott Klement <rpg400-l@xxxxxxxxxxxxxxxx>]

Rob,

I have no problem with him using pointers, provided that it's clear in 
his head how things work.  And I think it'd be great if he learns how to 
use pointers, and succeeds in getting the concepts straight in his head.

My caution was purely against writing pointer code that he doesn't 
understand (which is clearly the case at the moment) and then putting it 
into production somewhere.  He needs to clearly understand it.

As for him using a pointer to access the output of an API -- why on 
earth would you use the CLEAR op-code for that?!


rob@xxxxxxxxx wrote:
> Scott,
>
> I respectfully disagree with any general statements like removing pointer 
> logic from the program.
>
> And since he's using a data structure I am wondering if he's using 
> something like QUSPTRUS to process a list returned by some API.  I find 
> QUSPTRUS better than QUSRTVUS.
>
> Going back to QUSRTVUS would make as much sense as IBM dropping the ADD op 
> code and also dropping
> number = x + y;
> and making you do something like
> number = AddNumbers(x : y : ErrDS);
> Select;
> When ErrDS.ErrCode=NoError;
>   // ok
>   ProcessAddedUpNumbers(number);
> When ErrDS.ErrCode= TruncationError;
>   // Truncation occurred
> Other;
>   // generic error handler
> EndSl;
>
> He simply needs to consider whether or not he needs to initialize the data 
> structure at all, and if so, ensure that the basing pointer is set first.
>
> When using a data structure in list logic (like the API example above) I 
> wouldn't see a reason to ever initialize it.  The values would change upon 
> adding the increment to the pointer.  And if you checked the error code 
> upon the initial call to QUSPTRUS and didn't muff up the looping logic you 
> should be fine.
>
> Better to teach someone than to scare them off.
>
> Rob Berendt