Issue #10314 has been updated by nobu (Nobuyoshi Nakada).


olivierlacan (Olivier Lacan) wrote:
> Also, wasn't `def run(foo = self.foo)` a viable alternative to `def run(foo = foo())` to avoid circular reference warnings?

Unless `#foo` is private.

> Empty parens tend to trigger errors from tools like Rubocop.

Then you should report it to Rubocop, not here.

----------------------------------------
Bug #10314: Default argument lookup fails in Ruby 2.2 for circular shadowed variable names
https://bugs.ruby-lang.org/issues/10314#change-65356

* Author: lsegal (Loren Segal)
* Status: Open
* Priority: Normal
* Assignee: nobu (Nobuyoshi Nakada)
* Target version: 2.2.0
* ruby -v: 2.2.0-preview1
* Backport: 2.0.0: DONTNEED, 2.1: DONTNEED
----------------------------------------
The following code prints `nil` in Ruby 2.2.0-preview1 but worked in all previous version of Ruby back to 1.8.7:

~~~
class Foo
  def foo; "abc" end
 
  # this default param should resolve at runtime to the #foo method call
  def run(foo = foo)
    p foo # print shadowed local var defaulting to attr value
  end
end
 
puts "Testing #{RUBY_VERSION}:"
Foo.new.run

# Ruby 2.2.0-preview1
# => nil

# Ruby 1.x 2.x etc
# => "abc"
~~~

My guess is this is happening because "foo" in "foo = foo" is resolving to the argument variable "foo", which currently has the value of nil. It would be equivalent to setting "qux = qux" in a method body, which has been the expected behavior for a long time.

I understand that shadowing variables is something you should probably never do, but unfortunately this code was already written and working for quite a while, so I figured it would be wise to file a bug report for the following reasons:

1. This seems like a breaking change in Ruby 2.2.0-preview1 that was not announced in the changelog. My guess is this change may have been unintentional, but if it was we need a changelog entry at the very least.
2. If this is newly expected behavior, I wanted to chime in that I don't think it makes much sense. I can't think of any time when a user would expect the default value of a "foo = foo" argument to be the same foo argument itself. That would be tautologically nil. Arguably, it doesn't make much sense inside of a method body either when there is a shadowed method that could be called instead.



-- 
https://bugs.ruby-lang.org/

Unsubscribe: <mailto:ruby-core-request / ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>