Is there a security issue here? The slave application trusts the master application to have authenticated its user. But if the slave application can be called directly by an attacker, then authentication has been circumvented. How do you keep attackers from calling the slave application directly? Just through general access controls?
Would it make more sense to have the master query the user for authentication information (a cookie?) that will make the slave happy? Then, just pass that authentication information across to the slave untouched. The slave can then decide if that particular user is allowed access to that particular information. An outside attacker would still have to be able to pass along good authentication information if they managed to access the slave directly.
Twinotter 13:14, 17 July 2007 (PDT)
