class Prism::MultiTargetNode

^^^^^^
a, (b, c) = 1, 2, 3
Represents a multi-target expression.

def self.type

def self.type: () -> Symbol

class, but should be faster in a case statement or an array comparison.
Note that like #type, it will still be slower than using == for a single
splitting on the type of the node without having to do a long === chain.
Similar to #type, this method returns a symbol that you can use for 
def self.type
  :multi_target_node
end

def accept(visitor)

def accept: (Visitor visitor) -> void 
def accept(visitor)
  visitor.visit_multi_target_node(self)
end

def child_nodes

def child_nodes: () -> Array[nil | Node] 
def child_nodes
  [*lefts, rest, *rights]
end

def comment_targets

def comment_targets: () -> Array[Node | Location] 
def comment_targets
  [*lefts, *rest, *rights, *lparen_loc, *rparen_loc]
end

def compact_child_nodes

def compact_child_nodes: () -> Array[Node] 
def compact_child_nodes
  compact = []
  compact.concat(lefts)
  compact << rest if rest
  compact.concat(rights)
  compact
end

def copy(**params)

def copy: (**params) -> MultiTargetNode 
def copy(**params)
  MultiTargetNode.new(
    params.fetch(:lefts) { lefts },
    params.fetch(:rest) { rest },
    params.fetch(:rights) { rights },
    params.fetch(:lparen_loc) { lparen_loc },
    params.fetch(:rparen_loc) { rparen_loc },
    params.fetch(:location) { location },
  )
end

def deconstruct_keys(keys)

def deconstruct_keys: (Array[Symbol] keys) -> { lefts: Array[Node], rest: Node?, rights: Array[Node], lparen_loc: Location?, rparen_loc: Location?, location: Location } 
def deconstruct_keys(keys)
  { lefts: lefts, rest: rest, rights: rights, lparen_loc: lparen_loc, rparen_loc: rparen_loc, location: location }
end

def initialize(lefts, rest, rights, lparen_loc, rparen_loc, location)

def initialize: (Array[Node] lefts, Node? rest, Array[Node] rights, Location? lparen_loc, Location? rparen_loc, Location location) -> void 
def initialize(lefts, rest, rights, lparen_loc, rparen_loc, location)
  @newline = false
  @lefts = lefts
  @rest = rest
  @rights = rights
  @lparen_loc = lparen_loc
  @rparen_loc = rparen_loc
  @location = location
end

def inspect(inspector = NodeInspector.new)

def inspect(NodeInspector inspector) -> String 
def inspect(inspector = NodeInspector.new)
  inspector << inspector.header(self)
  inspector << "├── lefts: #{inspector.list("#{inspector.prefix}│   ", lefts)}"
  if (rest = self.rest).nil?
    inspector << "├── rest: ∅\n"
  else
    inspector << "├── rest:\n"
    inspector << rest.inspect(inspector.child_inspector("│   ")).delete_prefix(inspector.prefix)
  end
  inspector << "├── rights: #{inspector.list("#{inspector.prefix}│   ", rights)}"
  inspector << "├── lparen_loc: #{inspector.location(lparen_loc)}\n"
  inspector << "└── rparen_loc: #{inspector.location(rparen_loc)}\n"
  inspector.to_str
end

def lparen

def lparen: () -> String? 
def lparen
  lparen_loc&.slice
end

def rparen

def rparen: () -> String? 
def rparen
  rparen_loc&.slice
end

def type

def type: () -> Symbol

keys will use a jump table.
you can take advantage of the fact that case statements with all symbol
it uses a single integer comparison, but also because if you're on CRuby
can use for comparison. This is faster than the other approaches because
Instead, you can call #type, which will return to you a symbol that you

method calls, and/or array allocations.
these approaches are relatively slow because of the constant lookups,
case statement and doing `case node; when cls1; when cls2; end`. Both of
calling `[cls1, cls2].include?(node.class)` or putting the node into a
classes to see what kind of behavior to perform. Usually this is done by
Sometimes you want to check an instance of a node against a list of 
def type
  :multi_target_node
end