array_methods/joining_and_inspecting.rb

tests/_examples/array_methods/joining_and_inspecting.rb in the metorex repository.

# With no separator, or with nil for one, `$,` says what goes between.
p([1, 2, 3].join)
$, = "_"
p([1, 2, 3].join)
p([1, 2, 3].join(nil))
$, = nil

# An element is asked for `to_str`, then `to_ary`, then `to_s`, which is the
# order Ruby tries.
class Spells
  def to_str
    nil
  end

  def to_ary
    nil
  end

  def to_s
    "spelled"
  end
end
p([Spells.new].join)

# A join is written in the encoding the first piece is written in, widened by
# the first piece that is not all ASCII. Two pieces neither of which is ASCII
# and whose encodings differ cannot be joined at all.
p([].join.encoding)
p(["jp".encode("EUC-JP"), "utf8"].join.encoding)
begin
  ["\u{3042}", [0xFF].pack("C")].join
rescue EncodingError => refused
  p(refused.class.ancestors.include?(EncodingError))
end

# `inspect` asks what an element's own `inspect` answered for its `to_s` when
# that answer is no String, and never asks either one for `to_str`.
class Answers
  def inspect
    self
  end

  def to_s
    "answered"
  end
end
p([Answers.new].inspect)

# A symbol key prints without quotes only where it is named plainly enough to
# read back that way.
p({ a: 1, a!: 1, a?: 1 })
p({ "needs-quotes": 1 })
p({ "": 1 })

Output:

"123"
"1_2_3"
"1_2_3"
"spelled"
#<Encoding:US-ASCII>
#<Encoding:EUC-JP>
true
"[answered]"
{a: 1, a!: 1, a?: 1}
{"needs-quotes": 1}
{"": 1}