I usually get some mojibake if the output information (e.g. g.log.first.author.name) contains some Chinese character, just like '������' .

If I annotate the encoding code (lib.rb line 826 to 866), output would be correct (ruby 2.2.3).

  def command_lines(cmd, opts = [], chdir = true, redirect = '')
      cmd_op = command(cmd, opts, chdir)
    #   op = cmd_op.encode("UTF-8", "binary", {
      #     :invalid => :replace,
        # :undef => :replace
      # })
      cmd_op.split("\n")
  end

I have found an issue and a pull request to add this implementation. Does it a good choice to convert UTF-8 manually?