[ruby-core:123430] [Ruby Bug#21187] Strings concatenated with `\` getting frozen with literal hashes (PRISM only)

Issue #21187 has been updated by Earlopain (Earlopain _).


I think this should be marked for backport on 3.4. When doing so, cherry-pi=
ck changes to `prism_compile.c` from https://github.com/ruby/ruby/commit/a1=
403fb7cbd1fe0df97c932be9814c86081783dc and https://github.com/ruby/ruby/com=
mit/77aaa6ab0a475b8a57b1de4aa1f1210e4f85a803.

Then, pulling in the next prism release should make the fix effective. The =
first commit on its own with only `prism_compile.c` changes was benign but =
I can't tell if it would also be the case for the second commit. So I would=
 prefer to backport it all in one go and not start with something incomplet=
e, since my understanding is that prism gem changes should not be cherry-pi=
cked.

----------------------------------------
Bug #21187: Strings concatenated with `\` getting frozen with literal hashe=
s (PRISM only)
https://bugs.ruby-lang.org/issues/21187#change-114799

* Author: LocoDelAssembly (Hern=E1n Pereira)
* Status: Closed
* Assignee: prism
* ruby -v: ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [x86_64-linux]
* Backport: 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN
----------------------------------------
When the first elements of a literal hash are strings that are concatenated=
 with `\`, those elements are flagged with `PM_NODE_FLAG_STATIC_LITERAL` an=
d a special optimization that I believe was introduced in https://github.co=
m/ruby/ruby/commit/8080de04be8e99e71309745822a9d436cc4ae37c causes the stri=
ngs to be frozen.

Reproduction
=3D=3D=3D
test.rb
```
a =3D {
  a: 'one' \
     'two',
  b: 'three' \
     'four',
  c: 'five',
  d: 'six' \
     'seven'
}

b =3D {
  a: 'one',
  b: 'two' \
     'three'
}

puts "a =3D #{a.map { |k,v| {k =3D> v.frozen?} }}"
puts "b =3D #{b.map { |k,v| {k =3D> v.frozen?} }}"
```

With prism:
```
$ ruby test.rb                =20
a =3D [{a: true}, {b: true}, {c: false}, {d: false}]
b =3D [{a: false}, {b: false}]
```

With parse.y:
```
$ ruby --parser=3Dparse.y test.rb
a =3D [{a: false}, {b: false}, {c: false}, {d: false}]
b =3D [{a: false}, {b: false}]
```

(Notice `b` hash is unaffected in both parsers)

Not sure if this is just part of undefined behavior or this is indeed a bug=
. Assigning a string concatenated with `\` to a variable doesn't make it fr=
ozen, to the best of my knowledge this seems to be hash-specific.




--=20
https://bugs.ruby-lang.org/
______________________________________________
 ruby-core mailing list -- [e-mail]
 To unsubscribe send an email to [e-mail]
 ruby-core info -- https://ml.ruby-lang.org/mailman3/lists/ruby-core.ml.rub=
y-lang.org/