Re: EVAL(H) issue in V5R2

["Buck Calabro" <buck.calabro@xxxxxxxxxxxx>]

> EVAL(H)     stdMixO = (PSBTF1/shstdb) * shActB  gives 3.9993
> (all variables are defined as 15 4(dec)
> the same program running on V5R2   gives 3.9999
> Why V5R2 give a different result [from V4R5]?

Read up on the topic of intermediate precision in the RPG Reference
for all the details.  Understanding how expressions are evaluated is
generally a new idea for RPG programmers because of our history of one
calculation per line.  Because of that history, we declare our own
intermediate (work) variables.  When we use an expression, the
compiler has to create those work variables.  The size it chooses to
use depends on the variables used in the expression.  The maximum size
of a numeric variable changed between V4R5 and V5R2, thus altering the
compiler's decision and the final result.
  --buck