Could we from now on agree not to talk about (non)existing array elements,

Well we have an exists function, why shouldn't we talk about (non)existing array elements? :-)

Existence is an an important point to describe the behaviors of slicing, i.e. mostly returning undef for non-existing elements. (And I think what ikegami meant)

Let's express it in LRTs (Lanx's Rule of Thumb ;-)

1. Arrayslices (and hashslices) always return values for all requested elements. Not existing elements are represented as 'undef'.

  DB<1> @a=1..2
 
  DB<2> x @a[0,1]
0  1
1  2
  DB<3> x @a[0,3]
0  1
1  undef
  DB<4> x @a[2,3]
0  undef
1  undef  
  DB<5> @a=()
 
  DB<6> x @a[1,2]
0  undef
1  undef
[download]

2. Listlices work alike if and only if at least one of the requested elements exists. Otherwise it returns an empty list.

  DB<5> x (1,2)[0,1]
0  1
1  2
  DB<6> x (1,2)[0,3]
0  1
1  undef
  DB<7> x (1,2)[2,3]                # DIFFERENT!
  empty array
  DB<8> x ()[0..3]                  # DIFFERENT!
  empty array
  DB<9> x (undef,undef)[0,2]        
0  undef
1  undef
[download]

Note: Maybe the x command in the debugger should better say "empty list" instead of "empty array"....

Please correct me if I missed something with this "model" of slice behavior.

Cheers Rolf

Actually it should be defined what "empty" means. undef can be an existing element of a list!

  DB<1> p scalar ( ()=(undef,undef) )
2
[download]