From: "jasonmaxlee (Jason Lee) via ruby-core" <ruby-core@...>
Date: 2026-10-30T22:40:35+00:00
Subject: [ruby-core:127282] [Ruby Bug#18338] Encoding.default_external = Encoding::UTF_16BE may add a wrongly-encoded string to $LOADED_FEATURES

Issue #18338 has been updated by jasonmaxlee (Jason Lee).


I'd like to fix this. The cause is that on Unix the filesystem encoding is set from default_external (`Init_enc_set_filesystem_encoding` in localeinit.c, called from `enc_set_default_encoding`), so an ASCII-incompatible default_external such as UTF-16BE becomes the encoding of every path, including transcoder paths added to `$LOADED_FEATURES` by `String#encode`.
Proposal: when default_external is ASCII-incompatible, keep using the locale encoding (or ASCII-8BIT) for the filesystem encoding, since such an encoding can't describe a file path anyway. ASCII-compatible encodings keep following default_external as now. Then `test_object_inspect_external` can run again.
Pull request: https://github.com/ruby/ruby/pull/19628

----------------------------------------
Bug #18338: Encoding.default_external = Encoding::UTF_16BE may add a wrongly-encoded string to $LOADED_FEATURES
https://bugs.ruby-lang.org/issues/18338#change-119707

* Author: mame (Yusuke Endoh)
* Status: Open
* Backport: 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN
----------------------------------------
```
# coding: US-ASCII

Encoding.default_external = Encoding::UTF_16BE
"abc".encode(Encoding.default_external)

p $LOADED_FEATURES.last.encoding #=> #<Encoding:UTF-16BE>
p $LOADED_FEATURES.last
#=> "\u2F68\u6F6D\u652F\u6D61\u6D65\u2F77\u6F72\u6B2F\u7275\u6279\u2F6C\u6F63\u616C\u2F6C\u6962\u2F72\u7562\u792F\u332E\u312E\u302F\u7838\u365F\u3634\u2D6C\u696E\u7578\u2F65\u6E63\u2F74\u7261\u6E73\u2F75\u7466\u5F31\u365F\u3332\u2E73\x6F"
```

This weird string seems `"/home/mame/work/ruby/local/lib/ruby/3.1.0/x86_64-linux/enc/trans/utf_16_32.s\u0000o".force_encoding("UTF-16BE")`.

Note that the code may raise a "code converter not found" error depending on the length of install path (or build path?). Maybe it works if it is even due to UTF-16.


```
# works
mame$ /Users/mame/ruby2/exe/ruby --disable-gems test.rb
#<Encoding:UTF-16BE>
"\u2F55\u7365\u7273\u2F6D\u616D\u652F\u7275\u6279\u322F\u2E65\u7874\u2F78\u3836\u5F36\u342D\u6461\u7277\u696E\u3139\u2F65\u6E63\u2F74\u7261\u6E73\u2F73\u696E\u676C\u655F\u6279\u7465\u2E62\u756E\u646C\x65"
```

```
# raises an exception
mame$ /Users/mame/ruby22/exe/ruby --disable-gems test.rb
0ec: 0x0
test.rb:4:in `encode': code converter not found (US-ASCII to UTF-16BE) (Encoding::ConverterNotFoundError)
        from test.rb:4:in `<main>'
test.rb:4:in `encode': No such file or directory @ rb_check_realpath_internal - ������������������������������������������������������������������������������������������������������ (Errno::ENOENT)
        from test.rb:4:in `<main>'
```



-- 
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/lists/ruby-core.ml.ruby-lang.org/