Sure, that actually looks to be what other shells do anyhow.
Index: expr.c
===================================================================
RCS file: /cvs/src/bin/ksh/expr.c,v
retrieving revision 1.21
diff -u -p -r1.21 expr.c
--- expr.c 1 Jun 2009 19:00:57 -0000 1.21
+++ expr.c 27 Mar 2013 23:13:44 -0000
@@ -341,11 +341,17 @@ evalexpr(Expr_state *es, enum prec prec)
break;
case O_DIV:
case O_DIVASN:
- res = vl->val.i / vr->val.i;
+ if (vl->val.i == LONG_MIN && vr->val.i == -1)
+ res = LONG_MIN;
+ else
+ res = vl->val.i / vr->val.i;
break;
case O_MOD:
case O_MODASN:
- res = vl->val.i % vr->val.i;
+ if (vl->val.i == LONG_MIN && vr->val.i == -1)
+ res = 0;
+ else
+ res = vl->val.i % vr->val.i;
break;
case O_PLUS:
case O_PLUSASN:
On Wed, Mar 27, 2013 at 04:06:22PM -0700, Matthew Dempsky wrote:
> On Wed, Mar 27, 2013 at 3:56 PM, Nicholas Marriott
> <nicholas.marri...@gmail.com> wrote:
> > case O_DIV:
> > case O_DIVASN:
> > + if (vl->val.i == LONG_MIN && vr->val.i == -1)
> > + evalerr(es, ET_STR, "can't represent
> > result");
> > res = vl->val.i / vr->val.i;
> > break;
>
> I think LONG_MIN / -1 should still be LONG_MIN. We don't give errors
> for other overflow conditions.
