Collections

Birdie’s design makes commonly used algorithmic data structures directly available, without imports.

Design, not a released API

The collections below are proposed. The language implementation is still ahead of us.

Lists

[T] is a growable list. [int] is a list of integers, and [(int, int)] is a list of integer pairs.

• .size is the read-only element count.
• values[i] uses zero-based indexing.
• Negative and out-of-bounds indices cause a runtime error.
• .last returns the final element; an empty list causes a runtime error.

Lists can also support push and pop at the end.

Maps

Create an empty map with map(). Birdie should infer its key and value types from usage and request an annotation if there is not enough information.

seen = map()
seen[num] = i

key in seen tests membership. seen[key] looks up a value; a missing key causes a runtime error. Assignment inserts a new entry or replaces an existing one. .size counts entries.

A populated map can use a literal:

matching = {')': '(', ']': '[', '}': '{'}

Map iteration order and single-binding map iteration remain undecided.

Stacks

stack() creates a dedicated LIFO collection. Its element type is inferred from use.

opened = stack()
opened.push(char)

• push(value) adds an element to the top.
• pop() removes and returns the top element; an empty stack causes a runtime error.
• .size counts elements.
• if not opened: tests whether the stack is empty.

A stack has no indexing, arbitrary insertion, or sorting. Its restricted interface makes the intended algorithm explicit.

Further candidates

Heaps, queues/deques, sets, and counters are priorities for exploration. Sorted sets, union-find, tries, and graph/grid helpers are also candidates. Their APIs have not been defined here.