ruby-core

Issue #12861 has been updated by Shyouhei Urabe.


bug hit wrote:
> I think it would be better if super were always lexically scoped

Agreed but... What should then happen for inter-class block passing situation, like this?

```ruby
class Foo
  def self.foo
    'foo@foo'
  end
end

class Bar < Foo
  def self.bar(&block)
    define_singleton_method :foo, &block
  end
end

class Baz
  def self.foo
    Bar.bar do
      super
    end
  end
end

Baz.foo
```

----------------------------------------
Bug #12861: super in a block can be either lexically or dynamically scoped depending on how the block is invoked
https://bugs.ruby-lang.org/issues/12861#change-61037

* Author: bug hit
* Status: Open
* Priority: Normal
* Assignee: 
* ruby -v: ruby 2.3.1p112 (2016-04-26 revision 54768) [x86_64-linux]
* Backport: 2.1: UNKNOWN, 2.2: UNKNOWN, 2.3: UNKNOWN
----------------------------------------
```ruby
class Class1
  def self.foo
    'foo'
  end
  def self.method1
    'method1'
  end
end

class Class2 < Class1
  def self.foo
    bar do
      super()
    end
  end
  def self.bar(&block)
    a = block.()
    define_singleton_method :method1, &block
    b = send(:method1)
    c = block.()
    [a, b, c]
  end
end

p Class2.foo # ["foo", "method1", "foo"]
```

It doesn't seem like a good idea for a given language construct to be either lexically or dynamically scoped, depending on how its surrounding block is invoked (which is not visible at the point of definition).  I think it would be better if super were always lexically scoped, and a different keyword (dynamic_super) were always dynamically scoped



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

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