Slice objects with negative increment

[Alex Martelli]

Paul Hughett wrote:

> Alex Martelli <aleax at aleax.it> wrote:
> 
> : Paul Hughett wrote:
> 
> : Why ever would it be?  slice(j-L-1, i-L-1, -1) works per your stated
> : constraints and requirements for a sequence of length L, obviously.
> 
> Ah, ha!  It wasn't obvious to me that i-L-1 is an alias for i-1.  And

Isn't that the whole point of "negative indices alias to nonnegative
ones"?  It means that, for L>x>=0, x-L and x index the same item.

> it still seems bizarre to have to write that when I really mean i-1.

Unfortunately you don't _really_ mean that -- given said aliasing --
if the subtraction makes i from nonnegative to negative (i.e., if i
is exactly 0, in this case; easily generalizes to i-K for any given
K, of course).

> someone else's code.  Not that I can think of any solution, other than
> to explain this trick when slice objects are explained.

Explaining is surely good.  I think slicing can usefully be introduced
before slice-objects -- the latter are a tad harder because more general,
as they exist without specifically being tied to a given sequence (and
thus in particular to a given sequence _length_, which influences the
negative/nonnegative mapping).


Alex