News for hackers. Stuff that matters.
Re:Metasploit Rails 3 Remote Code Execution Hours Away by hjr3 on Thursday January 10, 2013 @12:07AM (#5035564)
I don't think defensive programming would have based on the current situation across a number of languages. Basically no serializer in any language PHP, Python, Rails, etc is completely safe. There are known exploits for all of them.

There is lots of magic in rails, but I bet there are a number of popular PHP and Python libraries/frameworks that are unserializing in an unsafe way.

Re:Metasploit Rails 3 Remote Code Execution Hours Away by ab on Thursday January 10, 2013 @12:14AM (#5035591)
There's an enormous difference between serializers that make any effort at all to be safe, and those like Ruby's YAML library, which make no effort. Python's YAML, for example, exposes a safe_load() method.

It's really criminally negligent that no such method exists in Ruby's YAML library.

Re:Metasploit Rails 3 Remote Code Execution Hours Away by ghostganz on Thursday January 10, 2013 @03:42AM (#5036076)
Python's Pickle lib had something similar to safe_load(), that they removed because it gave a false sense of security.
Re:Metasploit Rails 3 Remote Code Execution Hours Away by X-Istence on Thursday January 10, 2013 @08:39AM (#5036750)
If you are accepting pickled objects from a remote and using it ... you are an idiot.
Re:Metasploit Rails 3 Remote Code Execution Hours Away by vidarh on Thursday January 10, 2013 @06:04AM (#5036379)
Defensive programming in this case would be to not use a general purpose de-serializer for a format that is explicitly designed to allow serialising arbitrary objects....
Re:Metasploit Rails 3 Remote Code Execution Hours Away by gcr on Thursday January 10, 2013 @10:53AM (#5037543)
Serializers aren't inherently unsafe. Look at Lisp's (read) method, for example, which does not evaluate anything.