On 2/5/07, David Green wrote:
Then we wouldn't need * to count backwards, although it's still
useful to allow us to count past the end of an array. There are all
sorts of variations on this scheme, such as whether * is the last
element or the one after that, etc., or whether 0 should be the
first element or the last, and so on.
In some ways, I like not having a [0] index at all: programmers may
be used to counting from zero, but normal humans start with first,
second, third, ... third last, second last, ["first"] last, so it
would make sense to count:
@[EMAIL PROTECTED], 2, 3, ..., -3, -2, -1]
And then to count past the ends, something like:
[..., -*-2, -*-1, -*, 1, 2, 3, ..., -3, -2, -1, +*, *+1, *2, ...]
Or since * stands for everything, maybe that should be ..., *-2, *-1,
* , *+1, *+2, ...?
The thing I like best about this option is that it most clearly
preserves the notion of "*" as "everything"; *-1 is the element
before everything, and *+1 is the element after everything else.
If there is a way to work "modulated" (cyclic/wrap-around) arrays
into this, then we'd want to have a zero in there somewhere, although
it still wouldn't have to be first; instead count first, second,
third, ..., last but two, last but one, last:
[1, 2, 3, ..., -2, -1, 0]
(So, for a mod-4 array,
@[EMAIL PROTECTED],2,3,[EMAIL PROTECTED],6,7,[EMAIL PROTECTED],-2,-1,0] etc.?)
Yet another way (that avoids distinct +* and -*) is:
[..., -2,-1] @foo[0, 1, 2, ..., *-2, *-1] [+*, *+1, *+2, ...]
(or [..., -1, 0] @foo[1, 2, 3, ..., *-1, +*] [*+1, *+2, ...] or something.)
Then *+/-n always refers to the end point, and we can count
continuously across the front end of an array or the back.
-David
