rubinius

Comments and changes to this ticket

• 

  pluskid June 16th, 2008 @ 12:42 AM

  The code will fail on line 12 (map1[:a][:b] = 1) for me on both Ruby 1.8 and Rubinius. I'm not sure if I understood what you mean. But if I change the code from map1[][] = 1 to @map1[:a] = 1@, it will pass for both Ruby 1.8 and Rubinius.

• 

  Adam Gardiner June 16th, 2008 @ 02:21 PM

  Oops! This was a simplified example drawn from real code (rdebug), and when it gave the same error, I assumed it was correctly showing the problem in Rubinius...

  The correct test code should be as follows:

  class Foo
    def self.map1  # Defs
      @@map1 ||= {}
    end
  
    class << self  # SClass
      def map2     # Defn
        @@map2 ||= {}
      end
    end
  
    # Initialise the cvars
    map1[:a] ||= {}   
    map2[:a] ||= {}
    
    # Now try and use them
    map1[:a][:b] = 1  # works fine
    map2[:a][:b] = 1  # raises NoMethodError []= on NilClass
  end
  

  This will work fine on MRI, but the final assignment to map2 will fail on Rubinius.

• 

  Wilson Bilkovich June 25th, 2008 @ 10:20 AM

  • → Tag changed from “” to “bug”
  • → Milestone changed from “1.0 preview” to “0.10”
  • → State changed from “new” to “open”
• 

  Wilson Bilkovich July 10th, 2008 @ 03:37 PM

  • → Milestone changed from “0.10” to “0.11”