RE: RPG - conversion vs re-write -Reply -Reply

[David Morris <dmorris@xxxxxxxxxxxxx>]

>>> "Hans Boldt" <hboldt@VNET.IBM.COM> 09/12 7:53 am >>>


>>Even though the code won't compile  Mr. Cozzi's example would be a good
place to have OPTIONS(*VARTYPE: *VARSIZE), even without full
operational
descriptor support.  It would be nice if *VARSIZE worked for numerics.

>As I pointed out in another note, Bob's example won't work for an
even more fundamental reason.  Regarding *VARSIZE working for numerics,
that would be rather problematical.  *VARSIZE for char is straight-
forward in that you can refer only to what is passed using %SUBST.
For numerics, you would have to explicitly handle all possible formats
that you'd want to pass.  Perhaps using VALUE or CONST parameters
would be more appropriate?  After all, there are no format restrictions
on those parameter passing methods.

Mr. Cozzi's example is illustrative of what we are trying to accomplish. 
You can easily use a pointer to parse a numeric.  You cannot get a pointer to
a value or constant.  Even without the OPDESC support it would be better
than CALLB.

>>Your example works like a boolean value and is just
great...but...somewhere
in the manual (no page numbers in mine) it says that character fields are
treated as boolean when they are used as the result of a comparison.   Why
can't the IF make that assumption?  IE:  IF ErrFlg, should work but gives a
compile error unless coded as IF ErrFlg = *ON.  If it shouldn't work why
does
EVAL ErrFlg = A <> B work?

>That's a good question.  First, we decided that for the conditional
statements, the expression must return an indicator value.  The reason
should be clear.  Conditional statements perform one action if the value
of the expression is *ON, and another if the value is *OFF.  If the
expression could return some other value, the action would be undefined.

>Now, why can you assign an indicator value to a character field?  Well,
for the same reason that old RPG allows "MOVE *IN01 CHARF".  Indicators
are considered a subset of character data.  And so, they are fully
assignment compatible.

>A full indicator data type is coming in a yet to be announced release,
which will allow you to code "IF ErrFlag".

That would be nice.  I still think it should work if I can move fielda to 
*IN01. 

Thanks for the insight on why things work the way they do.

David Morris
+---
| 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
+---