From: "byroot (Jean Boussier) via ruby-core" <ruby-core@...>
Date: 2024-03-03T16:15:25+00:00
Subject: [ruby-core:117044] [Ruby master Bug#20307] `Hash#update` from compare_by_identity hash can have unfrozen string keys

Issue #20307 has been updated by byroot (Jean Boussier).


I don't understand the bug report nor the patch.

If you make a frozen copy of the provided key, you change the key. e.g.


```
cache = {}.compare_by_indentity
key = Random.bytes(5)

cache[key] = 1
p cache[key]

``` 

I'd expect `cache[key]` to return `1`, with what I understand of your patch, it wouldn't anymore.

Mutable hash keys aren't a new thing, it's only that regular hash have this special behavior for strings to keep a frozen copy of the string key, but they don't do that for any other type.

I don't see why identity hashes would need to copy this behavior.

----------------------------------------
Bug #20307: `Hash#update` from compare_by_identity hash can have unfrozen string keys
https://bugs.ruby-lang.org/issues/20307#change-107115

* Author: nobu (Nobuyoshi Nakada)
* Status: Open
* Backport: 3.0: REQUIRED, 3.1: REQUIRED, 3.2: REQUIRED, 3.3: REQUIRED
----------------------------------------
I don't think this behavior is expected.

```ruby
i = Hash.new.compare_by_identity
k = "a"
i[k] = 0
h = {}.update(i)
p h.compare_by_identity?  # => false
p h["a"]                  # => 0

k.upcase!
# `k` is still in `h`.
p h.keys.include?(k)      # => true

# but not found.
p((h.fetch(k) rescue $!)) # => #<KeyError: key not found: "A">
h["A"] = 1  
p h                       # => {"A"=>0, "A"=>1}
```

I expect `h` to still have `"a"=>0` entry.




-- 
https://bugs.ruby-lang.org/
 ______________________________________________
 ruby-core mailing list -- ruby-core@ml.ruby-lang.org
 To unsubscribe send an email to ruby-core-leave@ml.ruby-lang.org
 ruby-core info -- https://ml.ruby-lang.org/mailman3/postorius/lists/ruby-core.ml.ruby-lang.org/