Re: Define field as an indicator in D-specs

[Charles Wilt <charles.wilt@xxxxxxxxx>]

Undefined behavior usually is!

Charles


On Fri, Dec 6, 2013 at 7:32 PM, Jon Paris <jon.paris@xxxxxxxxxxxxxx> wrote:

> That is truly scary isn't it.
>
>
> On 2013-12-06, at 7:19 PM, Barbara Morris <bmorris@xxxxxxxxxx> wrote:
>
> > RPG isn't consistent though. Both DSPLYs are done in this program, so
> > it's both false and true.
> >
> > D fld             s             10a   inz('a')
> > C                   move      fld           *in10
> > C   10'cond t'      dsply
> > C  N10'ncond t'     dsply
> > C                   return
> >
> > Many languages specifically define that 0 is false and anything else is
> > true. But for RPG, it's undefined what will happen if an indicator
> > doesn't have the value '0' or '1'.
> >
> > With RPG, since it's undefined behaviour, the compiler is free to
> > generate code to compare the indicator to either '0' or '1' depending on
> > which makes more sense for that specific case.