Index from end and ranges

Summary#

^n is the position n elements before the end, so xs[^1] is the last element — on a string, an array and a list. a..b is the range from a up to (not including) b; either end may be left out, and either may be written from the end: s[1..3], s[2..], arr[..2], s[^3..]. A string's range is a substring, an array's is a new array.

Signature#

xs[^<int n>]                 // the element n before the end — xs[xs.Length - n] (Count on a list)
s[<start>..<end>]            // string → string, end exclusive; either bound may be ^n or omitted
arr[<start>..<end>]          // T[] → a new T[]

Description#

• The same answers as C#. ^1 is the last element and ^0 is one past it (reading it faults, as in C#). s[a..b] is s.Substring(a, b - a) — an out-of-range bound faults the way Substring does.
• Where they are read. Inside an indexer. C# also lets ^1 and 1..3 be stored as Index/Range values; Osy# has no such value, so outside [ ] they are refused by name with the indexer to write instead.
• A list takes ^n but not a range — as in C#, List<T> has no range indexer; copy a slice with xs.GetRange(start, count).
• The value is read once per use. The indexed value and each bound must be something that can be read again — a variable, a field, a literal — because the lowering reads the length and then the element. Anything else (a call) is refused with the fix: bind it to a local first.
• Precedence is C#'s: .. binds tighter than arithmetic, so a computed bound is parenthesised — s[1..(n - 1)].

Examples#

string Last(string s) => $"{s[^1]}";                        // "o" for "hello"
string Middle(string s) => s[1..^1];                        // "ell"
int[] Head(int[] xs) => xs[..2];                            // { 1, 2 } of { 1, 2, 3 }
int LastOf(List<int> xs) => xs[^1];

See also#

• List indexer — list[i], read and write
• Array literals — building the arrays and lists a range slices