Re: You know, I'm pretty sure it's zoned...

R. Bruce Hoffman, Jr. wrote:
>>> ...
>>>Oh, it's likely to continue all right, but not for the reason you just
>>>stated, although that's the party line.  ;-)
>>> ...

Then I wrote:
>> Please elucidate.  As far as I know, the reason the compiler continues
>> to use the same old rules for numeric formats is exactly the way
>> Doug explained it.  Whatever the reasons for the RPG compiler behaving
>> the way it did in the past, upward compatibility keeps it behaving the
>> same way now.  That's upward compatibility of source AND non-recompiled
>> objects.

Then Bruce wrote:
> Forward and backward compatability don't have anything to do with it. Non
> recompiled objects also don't have anything to do with it. It's the act
of
> compiling that causes this to appear.

Here's why not recompiling is an issue: Say you have 2 programs that both
have F-spec "definitions" only.  One calls the other with packed
parameters.
We invoke the "Bruce said we could" clause, and change the compiler.  One
program is recompiled but not the other.  Oooooooooooops.

We are not allowed to change the behaviour of compiled programs NOR the
meaning of existing source code.  Some deviations from this rule are
allowed,
very occasionally, but never for something as fundamental as this.

It's a very common error among RPG programmers to think that what appears
on an I spec is a definition, but it is not a definition; it is a
description
of a field in a input buffer.  Anything that is a source of common error
is probably "bad".  I don't disagree that this being the default behaviour
is not ideal, and if I could go back in time and ask those long ago RPG
compiler designers to reconsider their decision (read "slap them upside
their head"), I would.  But my dog ate my time machine, so I can't.

Barbara Morris