Bug #19448: [Hash] Using Set as default value

• Status: Rejected
• Assignee: -
• Target version: -
• ruby -v: ruby 3.1.3p185 (2022-11-24 revision 1a6b16756e) [x86_64-darwin22]
• Backport: 2.7: UNKNOWN, 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN

[bobanj (Boban Jovanoski)]

When using a hash and set is used as a default value, the keys method for the hash does not return expected values.
A workaround for this is provided in the attachment.

[byroot (Jean Boussier)]

• Status changed from Open to Rejected

This isn't a bug and is documented in Hash.new https://docs.ruby-lang.org/en/3.2/Hash.html#class-Hash-label-Default+Values

Note that the default value is used without being duplicated. It is not advised to set the default value to a mutable object:

synonyms = Hash.new([])
synonyms[:hello] # => []
synonyms[:hello] << :hi # => [:hi], but this mutates the default!
synonyms.default # => [:hi]
synonyms[:world] << :universe
synonyms[:world] # => [:hi, :universe], oops
synonyms.keys # => [], oops

To use a mutable object as default, it is recommended to use a default proc

[Hanmac (Hans Mackowiak)]

That's is known limitation when using Hash default value

Also your problem is that << would alter the same object you put into as default object

meaning:

h = Hash.new(Set.new)
h[:a] << 1
h[:b] << 2
p h[:c] #=> Set[1,2]

[sawa (Tsuyoshi Sawada)]

@byroot (Jean Boussier) (Jean Boussier), I think your comment is misleading if not irrelevant.

The issue's point is that, a key-value pair is not stored in the hash just by calling it. The relevant part in the example you cited is:

synonyms.keys # => [], oops

which is indeed not a bug.

To use a mutable object as default, it is recommended to use a default proc

is irrelevant/incomplete as a response to this issue because:

h = Hash.new{Set.new}

will not help. You have to explicitly store the key-value pair as in:

h = Hash.new{|hash, key| hash[key] = Set.new}

[byroot (Jean Boussier)]

@sawa (Tsuyoshi Sawada) I wasn't planning to answer the question myself, only to point the author to the detailed documentation of how Hash defaults works, so they can answer their own question.