pem-dev
[Top] [All Lists]

Re: New directions (was: Re; FYI)

1994-02-22 23:56:00
Before Bob shoots me down in flames ( :-) ), I will point out that I made a 
slight technical blunder in the handling of %-hack addresses.

On Wed, 23 Feb 1994, Mr Rhys Weatherley wrote:

Of course, there are nasties like jueneman%wotan(_at_)gte(_dot_)com, but they 
aren't 
impossible to handle:

      C=US, O=Internet, OU=com, OU=gte, OU=wotan, CN=jueneman

It is impossible to get a reversible mapping using this.  It should have 
been the following:

        C=US, O=Internet, OU=com, OU=gte, CN=jueneman(p)wotan

where "(p)" is the encoding of "%" in the RFC <-> X.400 stuff (RFC 1327 
is the latest version I have on me).

The Internet -> DN mapping algorithm is therefore:

1. Everything to the right of the last '@' is converted into a DN 
   according to the mechanism in my previous message.
2. Everything to the left of the last '@' is encoded using RFC 1327's 
   conventions and becomes the common name.
3. If there is no '@' in the Internet address or the characters '%' or 
   '!' appear to the right of the last '@', then complain to the user, 
   because the address either isn't a real Internet address or it has been 
   mangled.  If it really needs to be converted, then make the whole 
   DN a single CN field encoded according to RFC 1327.

The reverse mapping algorithm is:

1. If the DN does not begin with "C=??, O=Internet" then leave it as is, 
   as it probably is a conventional DN rather than an encoded Internet 
   address.
2. Convert the C and OU fields into an Internet domain name for the 
   right-hand side of the '@'.
3. Decode the CN field using RFC 1327 conventions for the left-hand side 
   of the '@'.

Aside: can the CN field be encoded using a T.61 string rather than a 
printable string?  If so, we probably don't need RFC 1327 at all.

Cheers,

Rhys.



<Prev in Thread] Current Thread [Next in Thread>