* Kragen Javier Sitaker <kra...@canonical.org> [2012-03-27 13:15]:
> There’s also the question of how to *write* a wider variety of
> invocations than `for (foo: somegen) { ... }`. I mean, to do
> Python’s:
>
> try:
> x = somegen.next()
> except StopIteration:
> y()
> else:
> z(x)
>
> you could do:
>
> sometype x;
> for (x: somegen) goto ok;
> y();
> if (0) {
> ok:
> z(x);
> }
>
> but that seems rather obscure!
>
> A reasonable, if not very C-like, syntax might be
>
> z(next(somegen) { y(); goto somewhere; });
>
> That’s only reasonable because you almost always want to handle the
> end of an iterator using some kind of nonlocal transfer of control,
> even if only a `break;`.
A reasonable and more C-like syntax seems very easy to devise by simply
crossing `if` with iterator-`for`:
next (x: somegen) { z(x) }
else { y() }
For the hell of it it might be even possible to use `if` itself, much
like you have made `for` carry several different semantics.
OTOH over in Perl 6-land they are discovering that it is better to
mostly not overload things (so `for` only does list iteration and there
is a new `loop (;;)` keyword for C-style iteration, and likewise many
other things that had multiple duties in Perl 5 have become multiple
separate things). So maybe it is actually better for both looping and
single-stepping iteration to to have dedicated keywords separate from
`for` and `if`.
Regards,
--
Aristotle Pagaltzis // <http://plasmasturm.org/>
--
To unsubscribe: http://lists.canonical.org/mailman/listinfo/kragen-discuss
