Re: [PATCH] Implement warning for wrong use of ||/&&

["Manuel López-Ibáñez" <lopezibanez at gmail dot com>]

Another nitpick. I think this fixes PR 17946 which already had a
patch. I didn't compare both solutions but it may be worthwhile to
take into account comments in bugzilla and the replies to the original
patch: http://gcc.gnu.org/ml/gcc-patches/2004-10/msg00990.html

(I keep looking for a suitable warning option for this).

Cheers,

Manuel.



On 12/02/07, Manuel López-Ibáñez <lopezibanez@gmail.com> wrote:
> On 12/02/07, Dirk Mueller <dmueller@suse.de> wrote:
> > +
> > +         switch (code)
> > +           {
> > +           case TRUTH_ANDIF_EXPR:
> > +           case TRUTH_ORIF_EXPR:
> > +            case TRUTH_AND_EXPR:
> > +            case TRUTH_OR_EXPR:
> > +             if (INTEGRAL_TYPE_P (TREE_TYPE (arg1))
> > +                   && !CONSTANT_CLASS_P (arg1)
> > +                   && !expl_eq_arg1
> > +                   && TREE_CODE (arg2) == INTEGER_CST
> > +                   && !integer_zerop (arg2))
> > +               {
> > +                 warning (OPT_Walways_true, "logical %<%s%> with non-zero constant "
> > +                          "will always evaluate as true",
> > +                          ((code == TRUTH_ANDIF_EXPR)
> > +                           || (code == TRUTH_AND_EXPR)) ? "&&" : "||");
> > +                 expl_eq_arg1 = expl_eq_arg2 = true;
> > +               }
>
> Just a nitpick. This will print "&&" or "||" even if the code contains
> "and" or "or". I think there is already a bug about this somewhere but
> I am not sure whether it is a general defect in GCC or something that
> must be fixed case by case.
>
> Cheers,
>
> Manuel.