RE: Strange behavior w/%editc()

On Wed, 27 Feb 2002, James Rich wrote:
>
>On Wed, 27 Feb 2002, Bob Cozzi (RPGIV) wrote:
>
>> There certainly needs to be:
>>
>> CharToDec()
>
>atoi()
> ...

James, atoi() is ok for a subset of the char-to-dec problems.
atoll() handles a few more (up to about 18 digits).  But there's
no complete solution for values with decimals like '1.23'.
You can call atof(), but that introduces errors through floating
point.  You can write your own procedure with a return type that's
ok for almost everything, say (30,9), but if you have a string
with 10 decimal places, you have to do some more work.

This is the kind of thing where a language built-in function
is necessary: a function that returns a different type based
on the parameters.  %dec() is an example; depending on the
parameters, it can return a value of any length and decimal
positions.

The char-to-dec function performed by MOVE/MOVEL isn't exactly
the same as what atoi does.  If you MOVE '1234N' to a
numeric result, the result gets -12345.  atoi would stop
at the N and return 1234.  To get -12345 from atoi, your string
would have to be '-12345'.

I imagine that any RPG version of char-to-dec would not
duplicate the MOVE behaviour which basically assumes that the
character value is actually a zoned value, but would instead
duplicate the atoi() and atof() behaviour (without the floating
point).

Barbara Morris