Re: [PATCH] Optimization of getarrvalue()

[Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>]

On Mon, Nov 21, 2016 at 4:30 AM, Jun T. <takimoto-j@xxxxxxxxxxxxxxxxx> wrote:
>
> Maybe, but I still don't understand, for example, the following:
>
> % a=()
> % nargs "${(@)a[0]}"
> 1
> % a=( foo )
> % nargs "${(@)a[0]}"
> 0

OK, I agree that one is strange.  However:

> % a=( foo bar )
> % nargs "${(@)a[3]}"
> 1
> % nargs "${(@)a[3,3]}"     # or "${(@)a[3,4]}"  "${(@)a[3,5]}" ...
> 0

Two subscripts separated by a comma mean something different than one
subscript.  a[3] is a scalar, so adding (@) doesn't matter; a scalar
in double quotes is the empty string.  a[3,3] is an array, and an
array with the (@) flag in double quotes resolves to nothing if the
array is empty.  The only magic here is the already-magic semantics of
double-quoting "${a[@]}".

> More generally, $a[i,j] with i=$#a+1 is treated specially.

??  It's not treated any differently than i=$#a+N for any N >=0, is it?