Re: BIFs requiring constants = frustrating

[CRPence <CRPbottle@xxxxxxxxx>]

  For numeric editing, a "data pointer" is what easily resolves the 
issue for variable precision & scale.  Last I recall reading [on one 
of the lists here], they are still not available as part of the RPG, 
but they are available via the MI.

  As for using the edit codes A & B, AFaIK the only difference is 
that the latter effects "blanks for a zero value" versus some amount 
of the character\digits zero & optionally a decimal separator 
effected by the former.  Thus I would expect a variation of the 
following may be able to rid of all the extra edtcde(B) processing; 
perhaps by testing each precision value separately to achieve 
identical rounded results as currently coded:

<code>

  If Amount=0;
   rtnAmount = '';
  Else;
   Select;
    When w_DecPos = 0;
      Amt0Dec = Amount;
      rtnAmount = %trim( %editc( Amt0Dec: 'A': *CURSYM ) );
    When w_DecPos = 1;
      Amt1Dec = Amount;
      rtnAmount = %trim( %editc( Amt1Dec: 'A': *CURSYM ) );
    When w_DecPos = 2;
      Amt2Dec = Amount;
      rtnAmount = %trim( %editc( Amt2Dec: 'A': *CURSYM ) );
   EndSl;
  EndIf;

</code>

  The edit mask API can be used to generate the mask for those 
three variations of the numeric scale [optionally to use stored 
masks; i.e. retrieved masks versus created during runtime] and then 
the address of which mask to use can be assigned just before the 
call to the edit API.  Although a reduction to one /edit invocation/ 
is achieved, that alone does not get the call into the assignment 
itself.

Regards, Chuck

Kurt Anderson wrote:
> %kds - number of keys to use has to be a constant
> %dec - length and decimal positions have to be constants
> %editc - edit code has to be constant
>
> I'm working with amounts that I'm writing to XML. They can have 
> differing decimal precision based on the client processed. I 
> thought, wonderful, I can create a procedure, pass in some 
> parameters, and simply insert those parms into a statement to
> format my return value. If only it were so easy. But why isn't
> it? What is the purpose of making BIF parms requiring constant
> values? Where is the flexibility in that? I suppose it may be
> easier on the RPG developers, but is it better for the language?
>
> Info about my example. Amounts are already rounded to the needed 
> precision. Also, we go out to 6 decimals, I cut the example down
> to only going out to two decimals.
>
> So in my example, instead of being able to code:
> rtnAmount = %trim( %editC( %dec( Amount: LEN_Amount: w_DecPos)
>                            : EditCode: *CurSym);
>
> I have to code:
>
> <code>
>
>  Select;
>    When w_DecPos = 0;
>      Amt0Dec = Amount;
>      Select;
>        When EditCode = 'A';
>          rtnAmount = %trim( %editc( Amt0Dec: 'A': *CURSYM ) );
>        When EditCode = 'B';
>          rtnAmount = %trim( %editc( Amt0Dec: 'B': *CURSYM ) );
>      EndSl;
>    When w_DecPos = 1;
>      Amt1Dec = Amount;
>      Select;
>        When EditCode = 'A';
>          rtnAmount = %trim( %editc( Amt1Dec: 'A': *CURSYM ) );
>        When EditCode = 'B';
>          rtnAmount = %trim( %editc( Amt1Dec: 'B': *CURSYM ) );
>      EndSl;
>    When w_DecPos = 2;
>      Amt2Dec = Amount;
>      Select;
>        When EditCode = 'A';
>          rtnAmount = %trim( %editc( Amt2Dec: 'A': *CURSYM ) );
>        When EditCode = 'B';
>          rtnAmount = %trim( %editc( Amt2Dec: 'B': *CURSYM ) );
>      EndSl;
>  EndSl;
>
> </code>
>
> If I'm missing something and there's an easier way to do what I'm
> doing, please let me know.