On 5/10/05, Aaron Crane <[EMAIL PROTECTED]> wrote:
> Damian Conway writes:
> > Just as $42 is a shorthand for $/[42], so too $<whatever> is a
> > shorthand for $/<whatever>.
>
> Isn't $42 a shorthand for $/[41] ?
>
> I think that having 1-based digit-variables but 0-based array indexes on
> $/ is really confusing; mistakes of this sort seem to confirm my view.
Yeah, I'm pretty sure they should be consolidated somehow. Of course
we could make $# zero-based. But there are other possibilities:
* $/[0] is like $&, the full text of the match
* $/[0] is the name of the matching rule, like in P::RD
The latter isn't applicable everywhere, but I have to say that it was
pretty handy. But I'm not sure I ever used its zeroth position to my
advantage, so it would probably be better of a method on $/.
The former case is actually quite elegant if we extend it to the new
matching semantics. Consider:
"foobarbaz" ~~ / (foo (bar)) (baz) /
Then:
$/[0] eq "foobarbaz";
$/[1] is a match object
$/[1][0] eq "foobar"
$/[1][1][0] eq "bar"
$/[2][0] eq "baz"
So now we have the strigification behavior on $/: it just returns its
zeroth element. The current meaning of $0 is now consistent and means
exactly the same thing it used to.
But I'm in the middle of a movie, so I haven't really thought through
the rest of it clearly. I think it breaks down somewhat in the
presence of quantifiers.
Luk
