On Monday, August 16, 2010 12:58:07 pm Barry A. Warsaw wrote:
> The one thing that looks weird to me is VRFY.  Since it never actually
> does verify the user, should we even claim to support the command?  Why
> not let subclasses claim support if they want to add it?

RFC 5321 section 4.5.1 states VRFY should be implemented in order to be 
considered an RFC 5321-compliant implementation.  But, in section 3.5.3 
paragraph 2 it states that if the actual verification was not performed but 
syntax was checked similar to RCPT, then the response code should be 252.

So my purposes for providing the plumbing for VRFY are:

1. Provide a basic, valid implementation to be as RFC 5321-compliant as 
possible.

2. Let users know the command is there so that it can be reimplemented as 
they build their solutions.