Re: /FREE RPG compiler errors

[Barbara Morris <bmorris@xxxxxxxxxx>]

"M. Lazarus" wrote:
> ...
>   I'm not advocating using it for new code, but when converting old code it
> would be *very* helpful.
> ...

Mark, the reason MOVE and its friends aren't included in /free is that
we didn't think the MOVE opcode was the best choice for nearly all the
places it's currently used.  In the places where MOVE is a natural
assignment, EVAL works the same.

/free is more intended for new code than for converted old code.  ("Old
code" is often code that has not been significantly changed since it was
converted from RPG III.)  We don't expect or want people to do any
wholesale conversions on their code to /free.

Here are a couple of reasons why I think it's not wise to convert "all"
existing code to /free:

1. As someone already pointed out in this thread, the bifs often don't
work exactly the same way as the opcodes.  Doing a conversion would
usually require some recoding and would always require a lot of testing.

2. The issue with the icky replacement for MOVE say by %SUBST and %EDITC
would be much less likely to come up with new code as with converted RPG
code.  Trying to convert code line-by-line rather than
section-by-section is where the problems arise.  Why is that MOVE trying
to put a numeric value into the second two bytes of A, anyway?  Maybe it
would be better if A was a data structure with a zoned subfield.  If
it's part of some general buildup of the value of A, maybe several MOVE
lines dealing with A could be replaced by one concatenation expression. 
(Similar to the way two MOVEAs and a loop can often be replaced with one
string expression.)