Issue #19985 has been updated by zverok (Victor Shepelev). Even funnier that bare `load` does support the convention: ``` $ ruby --disable-gems -r pathname -e "load Pathname.new('test')" -e:1:in `load': cannot load such file -- test (LoadError) ``` ---------------------------------------- Feature #19985: Support `Pathname` for `require` https://bugs.ruby-lang.org/issues/19985#change-105145 * Author: vo.x (Vit Ondruch) * Status: Open * Priority: Normal ---------------------------------------- It seems that RubyGems / Bundler `require` method overrides of accept `Pathname` as and argument ~~~ $ ruby -rpathname -e ' pa = Pathname.new("test") require pa ' <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require': cannot load such file -- test (LoadError) from <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require' from -e:3:in `<main>' ~~~ while plain Ruby require does not: ~~~ $ ruby --disable-gems -rpathname -e ' pa = Pathname.new("test") require pa ' -e:3:in `require': no implicit conversion of Pathname into String (TypeError) from -e:3:in `<main>' ~~~ This inconsistency is surprising. It seems that RubyGems / Bundler developers think [1] that Ruby `require` should also accept `Pathname`. ~~~ $ ruby -v ruby 3.2.2 (2023-03-30 revision e51014f9c0) [x86_64-linux] ~~~ [1]: https://github.com/rubygems/rubygems/pull/7128 -- https://bugs.ruby-lang.org/

Issue #19985 has been updated by zverok (Victor Shepelev).

That's not funny, it's expected, as load deals exclusively with paths, and require does not generally deal with paths.

Yeah, makes sense.

But in general, require does not deal with paths. I don't think we should add Pathname support to require.

Makes sense. Then, probably, it is up to RubyGems to remove this false conversion. ---------------------------------------- Feature #19985: Support `Pathname` for `require` https://bugs.ruby-lang.org/issues/19985#change-105147 * Author: vo.x (Vit Ondruch) * Status: Open * Priority: Normal ---------------------------------------- It seems that RubyGems / Bundler `require` method overrides of accept `Pathname` as and argument ~~~ $ ruby -rpathname -e ' pa = Pathname.new("test") require pa ' <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require': cannot load such file -- test (LoadError) from <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require' from -e:3:in `<main>' ~~~ while plain Ruby require does not: ~~~ $ ruby --disable-gems -rpathname -e ' pa = Pathname.new("test") require pa ' -e:3:in `require': no implicit conversion of Pathname into String (TypeError) from -e:3:in `<main>' ~~~ This inconsistency is surprising. It seems that RubyGems / Bundler developers think [1] that Ruby `require` should also accept `Pathname`. ~~~ $ ruby -v ruby 3.2.2 (2023-03-30 revision e51014f9c0) [x86_64-linux] ~~~ [1]: https://github.com/rubygems/rubygems/pull/7128 -- https://bugs.ruby-lang.org/

Issue #19985 has been updated by jeremyevans0 (Jeremy Evans). martinemde (Martin Emde) wrote in #note-5:

It is funny to me that you would say it’s not a path just because not all paths are valid. What else could it possibly be?

What the argument to `require` actually is is a feature name. That's why the argument to the method in the call-seq is `name` (not `filename` or `path`), and the first line of the `require` documentation mentions that it returns `true` or `false` depending on whether "the *feature* is already loaded".

Ruby docs for require:

...

If the filename neither resolves to an absolute path nor starts with ‘./’ or ‘../’, the file will be searched for in the library directories listed in $LOAD_PATH ($:). If the filename starts with ‘./’ or ‘../’, resolution is based on Dir.pwd.

https://ruby-doc.org/3.2.2/Kernel.html#method-i-require

Certainly the documentation for the `require` could be improved, because it implies that a filename that resolves to a path will be used, which is not generally the case as I showed previously. It's only the case for absolute paths or paths starting with `./` or `../`, and only if the path ends with a recognized file extension. In Ruby code, all valid `load` calls use paths. 99.9%+ of valid `require` calls do not use paths. I would guess that the majority of times that an argument starting with `/`, `./`, or `../` is passed to `require`, it probably still leaves off the file extension.

Walks like a duck, quacks like a duck, looks like a duck, referred to as a duck, lays duck eggs, but certain duck species not allowed => not duck.

It's a platypus, you shouldn't call it a duck just because it has a bill. :) ---------------------------------------- Feature #19985: Support `Pathname` for `require` https://bugs.ruby-lang.org/issues/19985#change-105150 * Author: vo.x (Vit Ondruch) * Status: Open * Priority: Normal ---------------------------------------- It seems that RubyGems / Bundler `require` method overrides of accept `Pathname` as and argument ~~~ $ ruby -rpathname -e ' pa = Pathname.new("test") require pa ' <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require': cannot load such file -- test (LoadError) from <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require' from -e:3:in `<main>' ~~~ while plain Ruby require does not: ~~~ $ ruby --disable-gems -rpathname -e ' pa = Pathname.new("test") require pa ' -e:3:in `require': no implicit conversion of Pathname into String (TypeError) from -e:3:in `<main>' ~~~ This inconsistency is surprising. It seems that RubyGems / Bundler developers think [1] that Ruby `require` should also accept `Pathname`. ~~~ $ ruby -v ruby 3.2.2 (2023-03-30 revision e51014f9c0) [x86_64-linux] ~~~ [1]: https://github.com/rubygems/rubygems/pull/7128 -- https://bugs.ruby-lang.org/

Issue #19985 has been updated by martinemde (Martin Emde). It seems like you’re suggesting that required features don’t use a path name based hierarchy for organization. I’m guessing you’re thinking of pathnames as something strictly tied to the filesystem, but Pathname is a useful class for manipulating any string that follows the slash separated, tree based navigation. We usually think of these as files on a file system, but ruby features are organized this way too and even if we broke them free of the file system, we’d still use this hierarchy. This is how you require specific subsets of features from Active Support, or how you require a feature that is not in load path by referring to where it can be found with relative pathname conventions that translate to filesystem locations. Paths, and therefore Pathnames, show how to get somewhere. You could use them for url paths, cache keys, filesystems or ruby features. The fact is that we use paths for ruby features. That string, no matter what you call it, is a path even if it is not a strict filesystem filename. Almost everyone thinks it’s a duck (pathname) and rubygems has been accepting ducks for years for 99% of Rubyists. In this case things that make it unique are so slight that most people are more confused that it doesn’t work than enlightened when they learn it’s not supposed to work (for some reason). This is not following the principal of least surprise. ---------------------------------------- Feature #19985: Support `Pathname` for `require` https://bugs.ruby-lang.org/issues/19985#change-105155 * Author: vo.x (Vit Ondruch) * Status: Open * Priority: Normal ---------------------------------------- It seems that RubyGems / Bundler `require` method overrides of accept `Pathname` as and argument ~~~ $ ruby -rpathname -e ' pa = Pathname.new("test") require pa ' <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require': cannot load such file -- test (LoadError) from <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require' from -e:3:in `<main>' ~~~ while plain Ruby require does not: ~~~ $ ruby --disable-gems -rpathname -e ' pa = Pathname.new("test") require pa ' -e:3:in `require': no implicit conversion of Pathname into String (TypeError) from -e:3:in `<main>' ~~~ This inconsistency is surprising. It seems that RubyGems / Bundler developers think [1] that Ruby `require` should also accept `Pathname`. ~~~ $ ruby -v ruby 3.2.2 (2023-03-30 revision e51014f9c0) [x86_64-linux] ~~~ [1]: https://github.com/rubygems/rubygems/pull/7128 -- https://bugs.ruby-lang.org/

Issue #19985 has been updated by mame (Yusuke Endoh). A little explanation. Actually, the builtin `Kernel#require` has accepted a Pathname object by calling #to_path. ``` $ cat test.rb puts "Hello" $ ruby -v --disable-gems -rpathname -e 'require Pathname("./test")' ruby 3.2.2 (2023-03-30 revision e51014f9c0) [x86_64-linux] Hello ``` However, if the file did not exist, it would unintentionally result in a TypeError. ``` $ ruby -v --disable-gems -rpathname -e 'require Pathname("./test2")' ruby 3.2.2 (2023-03-30 revision e51014f9c0) [x86_64-linux] -e:1:in `require': no implicit conversion of Pathname into String (TypeError) from -e:1:in `<main>' ``` Therefore, nobu has fixed the bug by commit:4329554f171fdb483cafa672df5f2a08741940c5. ``` $ ./local/bin/ruby -v --disable-gems -rpathname -e 'require Pathname("./test")' ruby 3.3.0dev (2023-11-06T08:58:47Z master 4329554f17) [x86_64-linux] Hello $ ./local/bin/ruby -v --disable-gems -rpathname -e 'require Pathname("./test2")' ruby 3.3.0dev (2023-11-06T08:58:47Z master 4329554f17) [x86_64-linux] -e:1:in `require': cannot load such file -- ./test2 (LoadError) from -e:1:in `<main>' ``` ---------------------------------------- Bug #19985: Confusing error message when nonexistent `Pathname` for `require` https://bugs.ruby-lang.org/issues/19985#change-105174 * Author: vo.x (Vit Ondruch) * Status: Closed * Priority: Normal * Backport: 3.0: REQUIRED, 3.1: REQUIRED, 3.2: REQUIRED ---------------------------------------- It seems that RubyGems / Bundler `require` method overrides of accept `Pathname` as and argument ~~~ $ ruby -rpathname -e ' pa = Pathname.new("test") require pa ' <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require': cannot load such file -- test (LoadError) from <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require' from -e:3:in `<main>' ~~~ while plain Ruby require does not: ~~~ $ ruby --disable-gems -rpathname -e ' pa = Pathname.new("test") require pa ' -e:3:in `require': no implicit conversion of Pathname into String (TypeError) from -e:3:in `<main>' ~~~ This inconsistency is surprising. It seems that RubyGems / Bundler developers think [1] that Ruby `require` should also accept `Pathname`. ~~~ $ ruby -v ruby 3.2.2 (2023-03-30 revision e51014f9c0) [x86_64-linux] ~~~ [1]: https://github.com/rubygems/rubygems/pull/7128 -- https://bugs.ruby-lang.org/

Issue #19985 has been updated by usa (Usaku NAKAMURA). note: also needs backporting b5c74d54887 ---------------------------------------- Bug #19985: Confusing error message when nonexistent `Pathname` for `require` https://bugs.ruby-lang.org/issues/19985#change-105203 * Author: vo.x (Vit Ondruch) * Status: Closed * Priority: Normal * Backport: 3.0: REQUIRED, 3.1: DONE, 3.2: REQUIRED ---------------------------------------- It seems that RubyGems / Bundler `require` method overrides of accept `Pathname` as and argument ~~~ $ ruby -rpathname -e ' pa = Pathname.new("test") require pa ' <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require': cannot load such file -- test (LoadError) from <internal:/usr/share/rubygems/rubygems/core_ext/kernel_require.rb>:85:in `require' from -e:3:in `<main>' ~~~ while plain Ruby require does not: ~~~ $ ruby --disable-gems -rpathname -e ' pa = Pathname.new("test") require pa ' -e:3:in `require': no implicit conversion of Pathname into String (TypeError) from -e:3:in `<main>' ~~~ This inconsistency is surprising. It seems that RubyGems / Bundler developers think [1] that Ruby `require` should also accept `Pathname`. ~~~ $ ruby -v ruby 3.2.2 (2023-03-30 revision e51014f9c0) [x86_64-linux] ~~~ [1]: https://github.com/rubygems/rubygems/pull/7128 -- https://bugs.ruby-lang.org/