[Nickle] Multiple definitions of variable names in same scope do not cause an error

[David M Guy]

But the following is a more common usage(imho):

/* BEGIN CODE */
int i;
for (i = 0; i < 10; i++)
    if (test(i))
        break;
if (i < 10)
    return i;
/* save 'i' in separate variable if necessary */
for (i = 0; i < 20; i++)
    if (test(i))
        break;
if (i < 20)
    return i;
/* END CODE */



And certainly this code does not suffer either way:

/* BEGIN CODE */
for (int i = 0; i < 10; i++)
    if (test(i))
        break;
/* END CODE */


Perhaps you could make a pragma, so the programmer could ask the
interpreter to enforce unique names? This would be greatly appreciated.

Thanks.
  - David Guy







On Mon, 10 Nov 2003, Bart Massey wrote:
> In message <Pine.GSO.4.58.0311101318390.28639 at sirius.cs.pdx.edu> you wrote:
> > I wrote some nickle test code in which I made a copy/paste error,
> > accidentally declaring a variable name twice, but the interpreter did not
> > raise an error.
> >
> > Question: Is this a bug or by design?
>
> By design. Each new definition of a name creates a new
> variable, and they are all scoped properly.  So it is
> normally harmless.
>
> OTOH, it's a bug that the manual doesn't make this clear.
>
> > Personally, I would prefer if the interpreter considered
> > this an error.
>
> Or perhaps a warning.  But keep in mind that there are some
> useful idioms that would be borken if we did this, e.g.
>
> 	int i;
> 	for (i = 0; i < 10; i++)
> 	    if (test(i))
> 	        break;
> 	if (i < 10)
> 	    return i;
>
> could not be repeated in a scope.
>
> It's a tough design call, and one on which we've waffled in
> the past.  We expect that most redefinitions are not errors,
> but it would take an empirical study to figure it out
> definitively.
>
> Thanks much for the report!
>
> 	Bart