This post originated from an RSS feed registered with Ruby Buzz by Eric Stewart. Original Post: Getting RuBlog Working Feed Title: Ponderings On Ruby Feed URL: http://blog.eric-stewart.com/category/programming-ruby.rss Feed Description: This is the Ruby related section of Eric Stewart's weblog. These entries will be the commentary of a long time Java/C++ programmer that started exploring Ruby in 2003. Latest Ruby Buzz Posts Latest Ruby Buzz Posts by Eric Stewart Latest Posts From Ponderings On Ruby

With a few minor changes to a custom rublog.cgi script, my new empty blog came up like a charm. I added a simple rdoc entry, and then a plain text entry, and they worked great!

But when I attempted to click on a specific Link, or a category, or the Syndicate link, up came the apache error message. A review of the logs pointed out that I Can’t modify frozen string. Ok, fine. A quick check of the code turned up this problem code:

  # Work out the request path, and make it local to out datadir. Ignore it
  # if it contains any '..' sequences

  def determine_request_path
    request_path = ""
    path = @environment['PATH_INFO']
    if path && path !~ /\.\./
      path.slice!(0) if path[0] == ?/
      request_path = path unless path.empty?
    end

    @is_rss = request_path.sub!(/index.rss$/, '') || @parameters.has_key?('rss')

    @path = request_path
  end

The error occurred when the code attempted to slice the path, since it was apparently frozen when fetched from the environment. The problem was fixed by changing the if block to:

    if path && path !~ /\.\./
      mutable_path = path.dup
                        mutable_path.slice!(0) if mutable_path[0] == ?/
                        request_path = mutable_path unless mutable_path.empty?
    end

For complete Ruby newbies like myself, the dup operation was necessary because it creates a new object with the same data, not state. If I had used path.clone instead of path.dup the frozen state would have been carried over.

Since Ruby 1.8.0 was released in August and the last RuBlog release 3f8 came out in March, I am not too shocked by all this.

Update: It turns out, the latest cvs versions of everything in the RubLog distribution work just fine with Ruby 1.8.0, so problem was already solved. It gave me a chance to hack a little ruby though!