Re: EVAL(H) issue in V5R2

Jon Paris wrote:
> 
> The extension to 31 digits in V5R2 could account for the difference since
> the DECPREC keyword only governs the maximum size of filed that you can
> define.  The precision of intermediate results was upped from 30 to 31 (and
> then to 63 or whatever) in V5R3.
> 

You can code any size of field you like, independent of DECPREC. 
DECPREC only controls the size of edited values.  So if you had
%EDITC(some expression : 'P'), the number of bytes in the result of
%EDITC would be constant, even though the length of the temporary used
to hold the expression might be larger.

In any given expression, DECPREC is thrown out if you have any operand
that is bigger than the DECPREC; for example, if you have DECPREC(30)
and you have a field declared with length 31, then DECPREC for that
particular statement goes to 31.  In V5R3, if you use any field declared
with between 32 and 63 digits, DECPREC for that statement goes to 63.