RFC 1506






Network Working Group                                      J. Houttuin
Request for Comments:  1506                           RARE Secretariat
RARE Technical Report: 6                                   August 1993


        A Tutorial on Gatewaying between X.400 and Internet Mail

Status of this Memo

   This memo provides information for the Internet community.  It does
   not specify an Internet standard.  Distribution of this memo is
   unlimited.

Introduction

   There are many ways in which X.400 and Internet (STD 11, RFC 822)
   mail systems can be interconnected. Addresses and service elements
   can be mapped onto each other in different ways. From the early
   available gateway implementations, one was not necessarily better
   than another, but the sole fact that each handled the mappings in a
   different way led to major interworking problems, especially when a
   message (or address) crossed more than one gateway. The need for one
   global standard on how to implement X.400 - Internet mail gatewaying
   was satisfied by the Internet Request For Comments 1327, titled
   "Mapping between X.400(1988)/ISO 10021 and RFC 822."

   This tutorial was produced especially to help new gateway managers
   find their way into the complicated subject of mail gatewaying
   according to RFC 1327. The need for such a tutorial can be
   illustrated by quoting the following discouraging paragraph from RFC
   1327, chapter 1: "Warning: the remainder of this specification is
   technically detailed. It will not make sense, except in the context
   of RFC 822 and X.400 (1988). Do not attempt to read this document
   unless you are familiar with these specifications."

   The introduction of this tutorial is general enough to be read not
   only by gateway managers, but also by e-mail managers who are new to
   gatewaying or to one of the two e-mail worlds in general. Parts of
   this introduction can be skipped as needed.

   For novice end-users, even this tutorial will be difficult to read.
   They are encouraged to use the COSINE MHS pocket user guide [14]
   instead.

   To a certain extent, this document can also be used as a reference
   guide to X.400 <-> RFC 822 gatewaying. Wherever there is a lack of
   detail in the tutorial, it will at least point to the corresponding
   chapters in other documents. As such, it shields the RFC 1327 novice



RARE Working Group on Mail and Messaging (WG-MSG)               [Page 1]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   from too much detail.

Acknowledgements

   This tutorial is heavily based on other documents, such as [2], [6],
   [7], [8], and [11], from which large parts of text were reproduced
   (slightly edited) by kind permission from the authors.

   The author would like to thank the following persons for their
   thorough reviews: Peter Cowen (Nexor), Urs Eppenberger (SWITCH), Erik
   Huizer (SURFnet), Steve Kille (ISODE Consortium), Paul Klarenberg
   (NetConsult), Felix Kugler (SWITCH), Sabine Luethi.

Disclaimer

   This document is not everywhere exact and/or complete in describing
   the involved standards. Irrelevant details are left out and some
   concepts are simplified for the ease of understanding. For reference
   purposes, always use the original documents.
































RARE Working Group on Mail and Messaging (WG-MSG)               [Page 2]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


Table of Contents

       1. An overview of relevant standards ........................   4
         1.1. What is X.400 ? ......................................   5
         1.2. What is an RFC ? .....................................   8
         1.3. What is RFC 822 ? ....................................   9
         1.4. What is RFC 1327 ? ...................................  11
       2. Service Elements .........................................  12
       3. Address mapping ..........................................  14
         3.1. X.400 addresses ......................................  15
           3.1.1. Standard Attributes ..............................  15
           3.1.2. Domain Defined Attributes ........................  17
           3.1.3. X.400 address notation ...........................  17
         3.2. RFC 822 addresses ....................................  19
         3.3. RFC 1327 address mapping .............................  20
           3.3.1. Default mapping ..................................  20
             3.3.1.1. X.400 -> RFC 822 .............................  20
             3.3.1.2. RFC 822 -> X.400 .............................  22
           3.3.2. Exception mapping ................................  23
             3.3.2.1. PersonalName and localpart mapping ...........  25
             3.3.2.2. X.400 domain and domainpart mapping ..........  26
               3.3.2.2.1. X.400 -> RFC 822 .........................  27
               3.3.2.2.2. RFC 822 -> X.400 .........................  28
         3.4. Table co-ordination ..................................  31
         3.5. Local additions ......................................  31
         3.6. Product specific formats .............................  32
         3.7. Guidelines for mapping rule definition ...............  34
       4. Conclusion ...............................................  35
       Appendix A. References ......................................  36
       Appendix B. Index  (Only available in the Postscript version)  37
       Appendix C. Abbreviations ...................................  37
       Appendix D. How to access the MHS Co-ordination Server ......  38
       Security Considerations .....................................  39
       Author's Address ............................................  39

















RARE Working Group on Mail and Messaging (WG-MSG)               [Page 3]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


1. An overview of relevant standards

   This chapter describes the history, status, future, and contents of
   the involved standards.

   There is a major difference between mail systems used in the USA and
   Europe. Mail systems originated mainly in the USA, where their
   explosive growth started as early as in the seventies. Different
   company-specific mail systems were developed simultaneously, which,
   of course, led to a high degree of incompatibility. The Advanced
   Research Projects Agency (ARPA), which had to use machines of many
   different manufacturers, triggered the development of the Internet
   and the TCP/IP protocol suite, which was later accepted as a standard
   by the US Department of Defense (DoD). The Internet mail format is
   defined in STD 11, RFC 822 and the protocol used for exchanging mail
   is known as the simple mail transfer protocol (SMTP) [1]. Together
   with UUCP and the BITNET protocol NJE, SMTP has become one of the
   main de facto mail standards in the US.

   Unfortunately, all these protocols were incompatible, which explains
   the need to come to an acceptable global mail standard.  CCITT and
   ISO began working on a norm and their work converged in what is now
   known as the X.400 Series Recommendations. One of the objectives was
   to define a superset of the existing systems, allowing for easier
   integration later on. Some typical positive features of X.400 are the
   store-and-forward mechanism, the hierarchical address space and the
   possibility of combining different types of body parts into one
   message body.

   In Europe, the mail system boom came later. Since there was not much
   equipment in place yet, it made sense to use X.400 as much as
   possible right from the beginning. A strong X.400 lobby existed,
   especially in West-Germany (DFN). In the R&D world, mostly EAN was
   used because it was the only affordable X.400 product at that time
   (Source-code licenses were free for academic institutions).

   At the moment, the two worlds of X.400 and SMTP are moving closer
   together. For instance, the United States Department of Defense, one
   of the early forces behind the Internet, has decided that future DoD
   networking should be based on ISO standards, implying a migration
   from SMTP to X.400. As an important example of harmonisation in the
   other direction, X.400 users in Europe have a need to communicate
   with the Internet. Due to the large traffic volume between the two
   nets it is not enough interconnecting them with a single
   international gateway.  The load on such a gateway would be too
   heavy. Direct access using local gateways is more feasible.

   Although the expected success of X.400 has been a bit disappointing



RARE Working Group on Mail and Messaging (WG-MSG)               [Page 4]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   (mainly because no good products were available), many still see the
   future of e-mail systems in the context of this standard.

   And regardless if in the long run X.400 will or will not take over
   the world of e-mail systems, SMTP cannot be neglected over the next
   ten years. Especially the simple installation procedures and the high
   degree of connectivity will contribute to a growing number of RFC 822
   installations in Europe and world-wide in the near future.

1.1. What is X.400 ?

   In October 1984, the Plenary Assembly of the CCITT accepted a
   standard to facilitate international message exchange between
   subscribers to computer based store-and-forward message services.
   This standard is known as the CCITT X.400 series recommendations
   ([16], from now on called X.400(84)) and happens to be the first
   CCITT recommendation for a network application. It should be noted
   that X.400(84) is based on work done in the IFIP Working Group 6.5,
   and that ISO at the same time was proceeding towards a compatible
   document. However, the standardisation efforts of CCITT and ISO did
   not converge in time (not until the 1988 version), to allow the
   publication of a common text.

   X.400(84) triggered the development of software implementing (parts
   of) the standard in the laboratories of almost all major computer
   vendors and many software houses. Similarly, public carriers in many
   countries started to plan X.400(84) based message systems that would
   be offered to the users as value added services. Early
   implementations appeared shortly after first drafts of the standard
   were published and a considerable number of commercial systems are
   available nowadays.

   X.400(84) describes a functional model for a Message Handling System
   (MHS) and associates services and protocols. The model illustrated in
   Figure 1.1. defines the components of a distributed messaging system.

   Users in the MHS environment are provided with the capability of
   sending and receiving messages. Users in the context of an MHS may be
   humans or application processes. The User Agent (UA) is a process
   that makes the services of the MTS available to the user. A UA may be
   implemented as a computer program that provides utilities to create,
   send, receive and perhaps archive messages. Each UA, and thus each
   user, is identified by a name (each user has its own UA).








RARE Working Group on Mail and Messaging (WG-MSG)               [Page 5]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


    -----------------------------------------------------------------
    |                user        user   Message Handling Environment|
    |                 |            |                                |
    |     ----------------------------------------------------------|
    |     |           |            |    Message Handling System    ||
    |     |         ----          ----                             ||
    |     |         |UA|          |UA|                             ||
    |     |         ----          ----                             ||
    |     |           |             |                              ||
    |     |       -------------------------------------------------||
    |     |       |   |             |   Message Transfer System   |||
    |     | ----  |  -----         -----                          |||
    |user-|-|UA|--|--|MTA|         |MTA|                          |||
    |     | ----  |  -----         -----                          |||
    |     |       |    \             /                            |||
    |     |       |     \           /                             |||
    |     |       |      \         /                              |||
    |     |       |       \       /                               |||
    |     |       |        \     /                                |||
    |     | ----  |         -----                                 |||
    |user-|-|UA|--|---------|MTA|                                 |||
    |     | ----  |         -----                                 |||
    |     |       -------------------------------------------------||
    |     ----------------------------------------------------------|
    -----------------------------------------------------------------
                    Fig. 1.1. X.400 functional model

   The Message Transfer system (MTS) transfers messages from an
   originating UA to a recipient UA. As implied by the Figure 1.1, data
   sent from UA to UA may be stored temporarily in several intermediate
   Message Transfer Agents (MTA), i.e., a store-and- forward mechanism
   is being used. An MTA forwards received messages to a next MTA or to
   the recipient UA.

   X.400(84) divides layer 7 of the OSI Reference Model into 2
   sublayers, the User Agent Layer (UAL) and the Message Transfer Layer
   (MTL) as shown in the Figure 1.2.

   The MTL is involved in the transport of messages from UA to UA, using
   one or several MTAs as intermediaries. By consequence, routing issues
   are entirely dealt with in the MTL. The MTL in fact corresponds to
   the postal service that forwards letters consisting of an envelope
   and a content. Two protocols, P1 and P3, are used between the MTL
   entities (MTA Entity (MTAE), and Submission and Delivery Entity
   (SDE)) to reliably transport messages. The UAL embodies  peer UA
   Entities (UAE), which interpret the content of a message and offer
   specific services to the application process.  Depending on the
   application to be supported on top of the MTL, one of several end-



RARE Working Group on Mail and Messaging (WG-MSG)               [Page 6]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   to-end protocols (Pc) is used between UAEs. For electronic mail,
   X.400(84) defines the protocol P2 as part of the InterPersonal
   Messaging Service (IPMS). Conceivably other UAL protocols may be
   defined, e.g., a protocol to support the exchange of electronic
   business documents.

       --------------------------------------------------------------
                   -----                          -----
       UA layer    |UAE|<----- P2, Pc ----------->|UAE|
                   -----                          -----
       --------------------------------------------------------------
                   ------          ------         -----
       MTA layer   |MTAE|<-- P1 -->|MTAE|<-- P3-->|SDE|
                   ------          ------         -----
       --------------------------------------------------------------
             xxxE = xxx Entity ;   SDE = Submission & Delivery Entity
       --------------------------------------------------------------
                           Fig. 1.2. X.400 Protocols

   The structure of an InterPersonal Message (IPM) can be visualised as
   in Figure 1.3. (Note that the envelope is not a part of the IPM; it
   is generated by the MTL).

                                                            Forwarded
    Message                                                 IP-message
    -                     ----------      --- ----------    -
    |  message-           |envelope|     /    | PDI    |    |
    |  content   IPM      ----------    /     ----------    |
    |  -         -        ----------   /      ----------    |
    |  |         |  IPM-  |heading |  /       |heading |    |
    |  |         |  body  ---------- /        ----------    |
    |  |         |  -     ----------/         ----------    |
    |  |         |  |     |bodypart|          |bodypart|    |
    |  |         |  |     ----------\         ----------    |
    |  |         |  |     ---------- \        ----------    |
    |  |         |  |     |bodypart|  \       |bodypart|    |
    |  |         |  |     ----------   \      ----------    |
    |  |         |  |          .        \                   |
    |  |         |  |          .         \                  |
    |  |         |  |     ----------      \   ----------    |
    |  |         |  |     |bodypart|       \  |bodypart|    |
    -  -         -  -     ----------        - ----------    -
                                      (PDI = Previous Delivery Info.)
                    Fig. 1.3. X.400 message structure

   An IPM heading contains information that is specific for an
   interpersonal message like 'originator', 'subject', etc. Each
   bodypart can contain one information type, text, voice or as a



RARE Working Group on Mail and Messaging (WG-MSG)               [Page 7]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   special case, a forwarded message. A forwarded message consists of
   the original message together with Previous Delivery Information
   (PDI), which is drawn from the original delivery envelope.

   Early experience with X.400(84) showed that the standard had various
   shortcomings. Therefore CCITT, in parallel with ISO, corrected and
   extended the specification during its 1984 to 1988 study period and
   produced a revised standard [17], which was accepted at the 1988
   CCITT Plenary Meeting [10].  Amongst others, X.400(88) differs from
   X.400(84) in that it defines a Message Store (MS), which can be seen
   as a kind of database for messages. An MS enables the end-user to run
   a UA locally, e.g., on a PC, whilst the messages are stored in the
   MS, which is co-located with the MTA. The MTA can thus always deliver
   incoming messages to the MS instead of to the UA. The MS can even
   automatically file incoming messages according to certain criteria.
   Other enhancements in the 88 version concern security and
   distribution lists.

1.2. What is an RFC ?

   The Internet, a loosely-organised international collaboration of
   autonomous, interconnected networks, supports host-to-host
   communication through voluntary adherence to open protocols and
   procedures defined by Internet Standards. There are also many
   isolated internets, i.e., sets of interconnected networks, that are
   not connected to the Internet but use the Internet Standards. The
   architecture and technical specifications of the Internet are the
   result of numerous research and development activities conducted over
   a period of two decades, performed by the network R&D community, by
   service and equipment vendors, and by government agencies around the
   world.

   In general, an Internet Standard is a specification that is stable
   and well-understood, is technically competent, has multiple,
   independent, and interoperable implementations with operational
   experience, enjoys significant public support, and is recognisably
   useful in some or all parts of the Internet.

   The principal set of Internet Standards is commonly known as the
   "TCP/IP protocol suite". As the Internet evolves, new protocols and
   services, in particular those for Open Systems Interconnection (OSI),
   have been and will be deployed in traditional TCP/IP environments,
   leading to an Internet that supports multiple protocol suites.

   The following organisations are involved in setting Internet
   standards.

   Internet standardisation is an organised activity of the Internet



RARE Working Group on Mail and Messaging (WG-MSG)               [Page 8]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   Society (ISOC). The ISOC is a professional society that is concerned
   with the growth and evolution of the world-wide Internet, with the
   way in which the Internet is and can be used, and with the social,
   political, and technical issues that arise as a result.

   The Internet Engineering Task Force (IETF) is the primary body
   developing new Internet Standard specifications. The IETF is composed
   of many Working Groups, which are organised into areas, each of which
   is co-ordinated by one or more Area Directors.

   The Internet Engineering Steering Group (IESG) is responsible for
   technical management of IETF activities and the approval of Internet
   standards specifications, using well-defined rules. The IESG is
   composed of the IETF Area Directors, some at-large members, and the
   chairperson of the IESG/IETF.

   The Internet Architecture Board (IAB) has been chartered by the
   Internet Society Board of Trustees to provide quality control and
   process appeals for the standards process, as well as external
   technical liaison, organizational oversight, and long-term
   architectural planning and research.

   Any individual or group (e.g., an IETF or RARE working group) can
   submit a document as a so-called Internet Draft. After the document
   is proven stable, the IESG may turn the Internet-Draft into a
   "Requests For Comments" (RFC). RFCs cover a wide range of topics,
   from early discussion of new research concepts to status memos about
   the Internet. All Internet Standards (STDs) are published as RFCs,
   but not all RFCs specify standards. Another sub-series of the RFCs
   are the RARE Technical Reports (RTRs).

   As an example, this tutorial also started out as an Internet-Draft.
   After almost one year of discussions and revisions it was approved by
   the IESG as an Informational RFC.

   Once a document is assigned an RFC number and published, that RFC is
   never revised or re-issued with the same number. Instead, a revision
   will lead to the document being re-issued with a higher number
   indicating that an older one is obsoleted.

1.3. What is RFC 822 ?

   STD 11, RFC 822 defines a standard for the format of Internet text
   messages. Messages consist of lines of text. No special provisions
   are made for encoding drawings, facsimile, speech, or structured
   text. No significant consideration has been given to questions of
   data compression or to transmission and storage efficiency, and the
   standard tends to be free with the number of bits consumed. For



RARE Working Group on Mail and Messaging (WG-MSG)               [Page 9]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   example, field names are specified as free text, rather than special
   terse codes.

   A general "memo" framework is used. That is, a message consists of
   some information in a rigid format (the 'headers'), followed by the
   main part of the message (the 'body'), with a format that is not
   specified in STD 11, RFC 822. It does define the syntax of several
   fields of the headers section; some of these fields must be included
   in all messages.

   STD 11, RFC 822 is used in conjunction with a number of different
   message transfer protocol environments (822-MTSs).

        - SMTP Networks: On the Internet and other TCP/IP networks,
          STD 11, RFC 822 is used in conjunction with two other
          standards: STD 10, RFC 821, also known as Simple Mail
          Transfer Protocol (SMTP) [1], and RFCs 1034 and 1035
          which specify the Domain Name System [3].

        - UUCP Networks: UUCP is the UNIX to UNIX CoPy protocol, which
          is usually used over dialup telephone networks to provide a
          simple message transfer mechanism.

        - BITNET: Some parts of Bitnet and related networks use STD
          11, RFC 822 related protocols, with EBCDIC encoding.

        - JNT Mail Networks: A number of X.25 networks, particularly
          those associated with the UK Academic Community, use the JNT
          (Joint Network Team) Mail Protocol, also known as Greybook.

   STD 11, RFC 822 is based on the assumption that there is an
   underlying service, which in RFC 1327 is called the 822-MTS service.
   The 822-MTS service provides three basic functions:

        1. Identification of a list of recipients.
        2. Identification of an error return address.
        3. Transfer of an RFC 822 message.

   It is possible to achieve 2) within the RFC 822 header.  Some 822-
   MTS protocols, in particular SMTP, can provide additional
   functionality, but as these are neither mandatory in SMTP, nor
   available in other 822-MTS protocols, they are not considered here.
   Details of aspects specific to two 822-MTS protocols are given in
   Appendices B and C of RFC 1327. An RFC 822 message consists of a
   header, and content which is uninterpreted ASCII text. The header is
   divided into fields, which are the protocol elements. Most of these
   fields are analogous to P2 heading fields, although some are
   analogous to MTS Service Elements.



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 10]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


1.4. What is RFC 1327 ?

   There is a large community using STD 11, RFC 822 based protocols for
   mail services, who will wish to communicate with users of the
   InterPersonal Messaging Service (IPMS) provided by X.400 systems, and
   the other way around. This will also be a requirement in cases where
   RFC 822 communities intend to make a transition to use X.400 (or the
   other way around, which also happens), as conversion will be needed
   to ensure a smooth service transition.

   The basic function of a mail gateway can be described as follows:
   receive a mail from one mail world, translate it into the formats of
   the other mail world and send it out again using the routing rules
   and protocols of that other world.

   Especially if a message crosses more than one gateway, it is
   important that all gateways have the same understanding of how things
   should be mapped. A simple example of what could go wrong otherwise
   is the following: A sends a message to B through a gateway and B's
   reply to A is being routed through another gateway.

   If the two gateways don't use the same mappings, it can be expected
   that the From and To addresses in the original mail and in the answer
   don't match, which is, to say the least, very confusing for the end-
   users (consider what happens if automated processes communicate via
   mail). More serious things can happen to addresses if a message
   crosses more than one gateway on its way from the originator to the
   recipient. As a real-life example, consider receiving a message from:

      Mary Plork 

   This is not what you would call user-friendly addressing.... RFC 1327
   describes a set of mappings that will enable a more transparent
   interworking between systems operating X.400 (both 84 and 88) and
   systems using RFC 822, or protocols derived from STD 11, RFC 822.

   RFC 1327 describes all mappings in term of X.400(88). It defines how
   these mappings should be applied to X.400(84) systems in its Appendix
   G.

   Some words about the history of RFC 1327: It started out in June
   1986, when RFC 987 defined for X.400(84) what RFC 1327 defines for
   X.400(84 and 88). RFC 1026 specified a number of additions and
   corrections to RFC 987. In December 1989, RFC 1138, which had a very
   short lifetime, was the first one to deal with X.400(88). It was
   obsoleted by RFC 1148 in March 1990. Finally, in May 1992, RFC 1327
   obsoleted all of its ancestors.



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 11]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


2. Service Elements

   Both RFC 822 and X.400 messages consist of certain service elements
   (such as 'originator' and 'subject'). As long as a message stays
   within its own world, the behaviour of such service elements is well
   defined. An important goal for a gateway is to maintain the highest
   possible service level when a message crosses the boundary between
   the two mail worlds.

   When a user originates a message, a number of services are available.
   RFC 1327 describes, for each service elements, to what extent it is
   supported for a recipient accessed through a gateway.  There are
   three levels of support:

        - Supported: Some of the mappings are quite straight-forward,
          such as '822.Subject:' <-> 'IPMS.Subject'.

        - Not supported: There may be a complete mismatch: certain
          service elements exist only in one of the two worlds (e.g.,
          interpersonal notifications).

        - Partially supported: When similar service elements exist in
          both worlds, but with slightly different interpretations,
          some tricks may be needed to provide the service over the
          gateway border.

   Apart from mapping between the service elements, a gateway must also
   map the types and values assigned to these service elements.  Again,
   this may in certain cases be very simple, e.g., 'IA5 -> ASCII'. The
   most complicated example is mapping address spaces. The problem is
   that address spaces are not something static that can be defined
   within RFC 1327. Address spaces change continuously, and they are
   defined by certain addressing authorities, which are not always
   parallel in the RFC 822 and the X.400 world. A valid mapping between
   two addresses assumes however that there is 'administrative
   equivalence' between the two domains in which the addresses exist
   (see also [13]).

   The following basic mappings are defined in RFC 1327. When going from
   RFC 822 to X.400, an RFC 822 message and the associated 822- MTS
   information is always mapped into an IPM (MTA, MTS, and IPMS
   Services). Going from X.400 to RFC 822, an RFC 822 message and the
   associated 822-MTS information may be derived from:

        - A Report (MTA, and MTS Services)

        - An InterPersonal Notification (IPN) (MTA, MTS, and IPMS
          services)



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 12]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993



        - An InterPersonal Message (IPM) (MTA, MTS, and IPMS services)

   Probes (MTA Service) have no equivalent in STD 10, RFC 821 or STD 11,
   RFC 822 and are thus handled by the gateway. The gateway's Probe
   confirmation should be interpreted as if the gateway were the final
   MTA to which the Probe was sent. Optionally, if the gateway uses RFC
   821 as an 822-MTS, it may use the results of the 'VRFY' command to
   test whether it would be able to deliver (or forward) mail to the
   mailbox under probe.

   MTS Messages containing Content Types other than those defined by the
   IPMS are not mapped by the gateway, and should be rejected at the
   gateway.

   Some basic examples of mappings between service elements are listed
   below.

    Service elements:

         RFC 822         X.400
         ------------------------------------------------
         Reply-To:       IPMS.Heading.reply-recipients
         Subject:        IPMS.Heading.subject
         In-Reply-To:    IPMS.Heading.replied-to-ipm
         References:     IPMS.Heading.related-IPMs
         To:             IPMS.Heading.primary-recipients
         Cc:             IPMS.Heading.copy-recipients

    Service element types:

         RFC 822         X.400
         ------------------------------------------------
         ASCII           PrintableString
         Boolean         Boolean

    Service element values:

         RFC 822         X.400
         ------------------------------------------------
         oh_dear         oh(u)dear
         False           00000000

   There are some mappings between service elements that are rather
   tricky and important enough to mention in this tutorial. These are
   the mappings of origination-related headers and some envelope fields:





RARE Working Group on Mail and Messaging (WG-MSG)              [Page 13]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


    RFC 822 -> X.400:

        - If Sender: is present, Sender: is mapped to
          IPMS.Heading.originator, and From: is mapped to
          IPMS.Heading.authorizing-users. If not, From: is mapped to
          IPMS.Heading.originator.

    X.400 -> RFC 822

        - If IPMS.Heading.authorizing-users is present,
          IPMS.Heading.originator is mapped to Sender:, and
          IPMS.Heading.authorizing-users is mapped to From: . If not,
          IPMS.Heading.originator is mapped to From:.

    Envelope attributes

        - RFC 1327 doesn't define how to map the MTS.OriginatorName and
          the MTS.RecipientName (often referred to as the P1.originator
          and P1.recipient), since this depends on which underlying 822-
          MTS is used. In the very common case that RFC 821 (SMTP) is
          used for this purpose, the mapping is normally as follows:

            MTS.Originator-name <->   MAIL FROM:
            MTS.Recipient-name  <->   RCPT TO:

   For more details, refer to RFC 1327, chapters 2.2 and 2.3.

3. Address mapping

   As address mapping is often considered the most complicated part of
   mapping between service element values, this subject is given a
   separate chapter in this tutorial.

   Both RFC 822 and X.400 have their own specific address formats. RFC
   822 addresses are text strings (e.g., "plork@tlec.nl"), whereas X.400
   addresses are binary encoded sets of attributes with values. Such
   binary addresses can be made readable for a human user by a number of
   notations; for instance:

        C=zz
        ADMD=ade
        PRMD=fhbo
        O=a bank
        S=plork
        G=mary

   The rest of this chapter deals with addressing issues and mappings
   between the two address forms in more detail.



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 14]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


3.1. X.400 addresses

   As already stated above, an X.400 address is modelled as a set of
   attributes. Some of these attributes are mandatory, others are
   optional. Each attribute has a type and a value, e.g., the Surname
   attribute has type IA5text, and an instance of this attribute could
   have the value 'Kille'. Attributes are divided into Standard
   Attributes (SAs) and Domain Defined Attributes (DDAs).

   X.400 defines four basic forms of addresses ([17], 18.5), of which
   the 'Mnemonic O/R Address' is the form that is most used, and is the
   only form that is dealt with in this tutorial. This is roughly the
   same address format as what in the 84 version was known as 'O/R
   names: form 1, variant 1' ([16] 3.3.2).

3.1.1. Standard Attributes

   Standard Attributes (SAs) are attributes that all X.400 installations
   are supposed to 'understand' (i.e., use for routing), for example:
   'country name', 'given name' or 'organizational unit'.  The most
   commonly used SAs in X.400(84) are:

        surName (S)
        givenName (G)
        initials (I*) (Zero or more)
        generationQualifier (GQ)
        OrganizationalUnits (OU1 OU2 OU3 OU4)
        OrganizationName (O)
        PrivateDomainName (PRMD)
        AdministrationDomainName (ADMD)
        CountryName (C)

   The combination of S, G, I* and GQ is often referred to as the
   PersonalName (PN).

   Although there is no hierarchy (of addressing authorities) defined by
   the standards, the following hierarchy is considered natural:

        PersonalName < OU4 < OU3 < OU2 < OU1 < O < P < A < C

   In addition to the SAs listed above, X.400(88) defines some extra
   attributes, the most important of which is

        Common Name (CN)

   CN can be used instead of or even together with PN. The problem in
   X.400(84) was that PN (S G I* GQ) was well suited to represent
   persons, but not roles and abstract objects, such as distribution



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 15]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   lists. Even though postmaster clearly is a role, not someone's real
   surname, it is quite usual in X.400(84) to address a postmaster with
   S=postmaster. In X.400(88), the same postmaster would be addressed
   with CN=postmaster .

   The attributes C and ADMD are mandatory (i.e., they must be present),
   and may not be empty. At least one of the attributes PRMD, O, OU, PN
   and CN must be present.

   PRMD and ADMD are often felt to be routing attributes that don't
   really belong in addresses. As an example of how such address
   attributes can be used for the purpose of routing, consider two
   special values for ADMD:

        - ADMD=0; (zero) should be interpreted as 'the PRMD in this
          address is not connected to any ADMD'

        - ADMD= ; (single SPACE) should be interpreted as 'the PRMD in
          this address is reachable via any ADMD in this country'. It
          is expected that ISO will express this 'any' value by means
          of a missing ADMD attribute in future versions of MOTIS.
          This representation can uniquely identify the meaning 'any',
          as a missing or empty ADMD field as such is not allowed.

   Addresses are defined in X.400 using the Abstract Syntax Notation One
   (ASN.1). X.409 defines how definitions in ASN.1 should be encoded
   into binary format. Note that the meaning, and thus the ASN.1
   encoding, of a missing attribute is not the same as that of an empty
   attribute. In addressing, this difference is often represented as
   follows:

        - PRMD=; means that this attribute is present in the address,
          but its value is empty. Since this is not very useful, it's
          hardly ever used. The only examples the author knows of
          were caused by mail managers who should have had this
          tutorial before they started defining their addresses :-)

        - PRMD=@; means that this attribute is not present in the
          address. {NB. This is only necessary if an address notation
          (see 3.1.3) requires that every single attribute in the
          hierarchy is somehow listed. Otherwise, a missing attribute
          can of course be represented by simply not mentioning it.
          This means that this syntax is mostly used in mapping rules,
          not by end users.}

   Addresses that only contain SAs are often referred to as Standard
   Attribute Addresses (SAAs).




RARE Working Group on Mail and Messaging (WG-MSG)              [Page 16]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


3.1.2. Domain Defined Attributes

   Domain Defined Attributes (DDAs) can be used in addition to Standard
   Attributes. An instance of a DDA consists of a type and a value. DDAs
   are meant to have a meaning only within a certain context (originally
   this was supposed to be the context of a certain management domain,
   hence the name DDA), such as a company context.

   As an example, a company might want to define a DDA for describing
   internal telephone numbers: DDA type=phone value=9571.

   A bit tricky is the use of DDAs to encode service element types or
   values that are only available on one side of a service gateway.  The
   most important examples of such usage are defined in:

       RFC 1327 (e.g., DDA type=RFC-822 value=u(u)ser(a)isode.com)

       RFC 1328 (e.g., DDA type=CommonName value=mhs-discussion-list)

   Addresses that contain both SAs and DDAs are often referred to as DDA
   addresses.

3.1.3. X.400 address notation

   X.400 only prescribes the binary encoding of addresses, it doesn't
   standardise how such addresses should be written on paper or what
   they should look like in a user interface on a computer screen.
   There exist a number of recommendations for X.400 address
   representation though.

  - JTC proposed an annex to CCITT Rec. F.401 and ISO/IEC 10021-2,
    called 'Representation of O/R addresses for human usage'. According
    to this proposal, an X.400 address would look as follows:

    G=jo; S=plork; O=a bank; OU1=owe; OU2=you; P=fhbo; A=ade; C=zz

      Note that in this format, the order of O and the OUs is exactly
      the opposite of what one would expect intuitively (the attribute
      hierarchy is increasing from left to right, except for the O and
      OUs, where it's right to left. The reasoning behind this is that
      this sequence is following the example of a postal address). This
      proposal has been added (as a recommendation) to the 1992 version
      of the standards.

  - Following what was originally used in the DFN-EAN software, most
    EAN versions today use an address representation similar to the JTC
    proposal, with a few differences:




RARE Working Group on Mail and Messaging (WG-MSG)              [Page 17]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


            - natural ordering for O and OUs
            - no numbering of OUs.
            - allows writing ADMD and PRMD instead of A and P

    The address in the example above could, in EAN, be represented as:

    G=jo; S=plork; OU=you; OU=owe; O=a bank; PRMD=fhbo; ADMD=ade; C=zz

    This DFN-EAN format is still often referred to as _the_ 'readable
    format'.

  - The RARE Working Group on Mail and Messaging, WG-MSG, has made a
    recommendation that is very similar to the DFN-EAN format, but with
    the hierarchy reversed. Further, ADMD and PRMD are used instead of
    A and P. This results in the address above being represented as:

    C=zz; ADMD=ade; PRMD=fhbo; O=a bank; OU=owe; OU=you; S=plork; G=jo

    This format is recognised by most versions of the EAN software. In
    the R&D community, this is one of the most popular address
    representations for business cards, letter heads, etc. It is also
    the format that will be used for the examples in this tutorial.
    (NB. The syntax used here for describing DDAs is as follows:
    DD.'type'='value', e.g., DD.phone=9571)

  - RFC 1327 defines a slash separated address representation:

    /G=jo/S=plork/OU=you/OU=owe/O=a bank/P=fhbo/A=ade/C=zz/

    Not only is this format used by the PP software, it is also
    widespread for business cards and letter heads in the R&D
    community.

  - RFC 1327 finally defines yet another format for X.400 _domains_
    (not for human users):

    OU$you.OU$owe.O$a bank.P$fhbo.A$ade.C$zz

    The main advantage of this format is that it is better machine-
    parseble than the others, which also immediately implies its main
    disadvantage: it is barely readable for humans. Every attribute
    within the hierarchy should be listed, thus a missing attribute
    must be represented by the '@' sign
    (e.g., $a bank.P$@.A$ade.C$zz).

  - Paul-Andre Pays (INRIA) has proposed a format that combines the
    readability of the JTC format with the parsebility of the RFC 1327
    domain format. Although a number of operational tools within the GO-



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 18]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


    MHS community are already based on (variants of) this proposal, its
    future is still uncertain.

3.2. RFC 822 addresses

   An RFC 822 address is an ASCII string of the following form:

        localpart@domainpart

    "domainpart" is sub-divided into

    domainpart = sdom(n).sdom(n-1)....sdom(2).sdom(1).dom

    "sdom" stands for "subdomain", "dom" stands for "top-level-domain".

    "localpart" ;is normally a login name, and thus typically is a
    surname or an abbreviation for this. It can also designate a local
    distribution list.

    The hierarchy (of addressing authorities) in an RFC 822 address is
    as follows:

        localpart < sdom(n) < sdom(n-1) <...< dom

    Some virtual real-life examples:

        joemp@tlec.nl
        tsjaka.kahn@walhalla.diku.dk
        a13_vk@cs.rochester.edu

    In the above examples, 'nl', 'dk', and 'edu' are valid,
    registered, top level domains. Note that some networks that have
    their own addressing schemes are also reachable by way of 'RFC
    822-like' addressing. Consider the following addresses:

        oops!user          (a UUCP address)
        V13ENZACC@CZKETH5A (a BITNET address)

    These addresses can be expressed in RFC 822 format:

        user@oops.uucp
        V13ENZACC@CZKETH5A.BITNET

   Note that the domains '.uucp' and '.bitnet' have no registered
   Internet routing.  Such addresses must always be routed to a gateway
   (how this is done is outside the scope of this tutorial).

   As for mapping such addresses to X.400, there is no direct mapping



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 19]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   defined between X.400 on the one hand and UUCP and BITNET on the
   other, so they are normally mapped to RFC 822 style first, and then
   to X.400 if needed.

3.3. RFC 1327 address mapping

   Despite the difference in address formats, the address spaces defined
   by RFC 822 and X.400 are quite similar. The most important parallels
   are:

        - both address spaces are hierarchical
        - top level domains and country codes are often the same
        - localparts and surnames are often the same

   This similarity can of course be exploited in address mapping
   algorithms. This is also done in RFC 1327 (NB only in the exception
   mapping algorithm. See chapter 3.3.2).

   Note that the actual mapping algorithm is much more complicated than
   shown below. For details, see RFC 1327, chapter 4.

3.3.1. Default mapping

   The default RFC 1327 address mapping can be visualised as a function
   with input and output parameters:

          address information of the gateway performing the mapping
                                      |
                                      v
                             +-----------------+
        RFC 822 address <--->| address mapping | <---> X.400 address
                             +-----------------+

   I.e., to map an address from X.400 to RFC 822 or vice versa, the only
   extra input needed is the address information of the local gateway.

3.3.1.1. X.400 -> RFC 822

   There are two kinds of default address mapping from X.400 to RFC 822:
   one to map a real X.400 address to RFC 822, and another to decode an
   RFC 822 address that was mapped to X.400 (i.e., to reverse the
   default RFC 822 -> X.400 mapping).

   To map a real X.400 address to RFC 822, the slash separated notation
   of the X.400 address (see chapter 3.1.) is mapped to 'localpart', and
   the local RFC 822 domain of the gateway that performs the mapping is
   used as the domain part. As an example, the gateway 'gw.switch.ch'
   would perform the following mappings:



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 20]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


        C=zz; ADMD=ade; PRMD=fhbo; O=tlec; S=plork; ->
        /C=zz/ADMD=ade/PRMD=fhbo/O=tlec/S=plork/@gw.switch.ch

        C=zz; ADMD=ade; PRMD=fhbo; O=a bank; S=plork->
        "/C=zz/ADMD=ade/PRMD=fhbo/O=a bank/S=plork/"@gw.switch.ch

   The quotes in the second example are mandatory if the X.400 address
   contains spaces, otherwise the syntax rules for the RFC 822 localpart
   would be violated.

   This default mapping algorithm is generally referred to as 'left-
   hand-side encoding'.

   To reverse the default RFC 822 -> X.400 mapping (see chapter
   3.3.1.2): if the X.400 address contains a DDA of the type RFC-822,
   the SAs can be discarded, and the value of this DDA is the desired
   RFC 822 address (NB. Some characters in the DDA value must be decoded
   first. See chapter 3.3.1.2.). For example, the gateway

        DD.RFC-822=bush(a)dole.us; C=nl; ADMD=tlec; PRMD=GW
        ->
        bush@dole.us

3.3.1.2. RFC 822 -> X.400

   There are also two kinds of default address mapping from RFC 822 to
   X.400: one to map a real RFC 822 address to X.400, and another to
   decode an X.400 address that was mapped to RFC 822 (i.e., to reverse
   the default X.400 -> RFC 822 mapping).

   To map a real RFC 822 address to X.400, the RFC 822 address is
   encoded in a DDA of type RFC-822 , and the SAs of the local gateway
   performing the mapping are added to form the complete X.400 address.
   This mapping is generally referred to as 'DDA mapping'. As an
   example, the gateway 'C=nl; ADMD=tlec; PRMD=GW' would perform the
   following mapping:

        bush@dole.us  ->
        DD.RFC-822=bush(a)dole.us; C=nl; ADMD=tlec; PRMD=GW

   As for the encoding/decoding of RFC 822 addresses in DDAs, it is
   noted that RFC 822 addresses may contain characters (@ ! % etc.) that
   cannot directly be represented in a DDA. DDAs are of the restricted
   character set type 'PrintableString', which is a subset of IA5
   (=ASCII). Characters not in this set need a special encoding. Some
   examples (For details, refer to RFC 1327, chapter 3.4.):





RARE Working Group on Mail and Messaging (WG-MSG)              [Page 21]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


        100%name@address   -> DD.RFC-822;=100(p)name(a)address
        u_ser!name@address -> DD.RFC-822;=u(u)ser(b)name(a)address

   To decode an X.400 address that was mapped to RFC 822: if the RFC 822
   address has a slash separated representation of a complete X.400
   mnemonic O/R address in its localpart, that address is the result of
   the mapping. As an example, the gateway 'gw.switch.ch' would perform
   the following mapping:

        /C=zz/ADMD=ade/PRMD=fhbo/O=tlec/S=plork/G=mary/@gw.switch.ch
        ->
        C=zz; ADMD=ade; PRMD=fhbo; O=tlec; S=plork; G=mary

3.3.2. Exception mapping according to mapping tables

   Chapter 3.3.1. showed that it is theoretically possible to use RFC
   1327 with default mapping only. Although this provides a very simple,
   straightforward way to map addresses, there are some very good
   reasons not to use RFC 1327 this way:

        - RFC 822 users are used to writing simple addresses of the
          form 'localpart@domainpart'. They often consider X.400
          addresses, and thus also the left-hand-side encoded
          equivalents, as unnecessarily long and complicated. They
          would rather be able to address an X.400 user as if she had a
          'normal' RFC 822 address. For example, take the mapping

            C=zz; ADMD=ade; PRMD=fhbo; O=tlec; S=plork;     ->
            /C=zz/ADMD=ade/PRMD=fhbo/O=tlec/S=plork/@gw.switch.ch

          from chapter 3.3.1.1. RFC 822 users would find it much more
          'natural' if this address could be expressed in RFC 822 as:

            plork@tlec.fhbo.ade.nl

        - X.400 users are used to using X.400 addresses with SAs only.
          They often consider DDA addresses as complicated, especially
          if they have to encode the special characters, @ % ! etc,
          manually. They would rather be able to address an RFC 822
          user as if he had a 'normal' X.400 address. For example, take
          the mapping

            bush@dole.us
            ->
            DD.RFC-822=bush(a)dole.us;
            C=nl; ADMD= ; PRMD=tlec; O=gateway

          from chapter 3.3.1.2. X.400 users would find it much more



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 22]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


          'natural' if this address could be expressed in X.400 as:

            C=us; ADMD=dole; S=bush

        - Many organisations are using both RFC 822 and X.400
          internally, and still want all their users to have a simple,
          unique address in both mail worlds. Note that in the default
          mapping, the mapped form of an address completely depends on
          which gateway  performed the mapping. This also results in a
          complication of a more technical nature:

        - The tricky 'third party problem'. This problem need not
          necessarily be understood to read the rest of this chapter.
          If it looks too complicated, please feel free to skip it
          until you are more familiar with the basics.

          The third party problem is a routing problem caused by
          mapping. As an example for DDA mappings (the example holds
          just as well for left-hand-side encoding), consider the
          following situation (see Fig. 3.1.): RFC 822 user X in
          country A sends a message to two recipients: RFC 822 user Y,
          and X.400 user Z, both in country B:

            From: X@A
            To:   Y@B ,
                  /C=B/.../S=Z/@GW.A

          Since the gateway in country A maps all addresses in the
          message, Z will see both X's and Y's address as DDA-encoded
          RFC 822 addresses, with the SAs of the gateway in country A:

            From: DD.RFC-822=X(a)A; C=A;....;O=GW
            To:   DD.RFC-822=Y(a)B; C=A;....;O=GW ,
                  C=B;...;S=Z

















RARE Working Group on Mail and Messaging (WG-MSG)              [Page 23]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993



            |       ------------         ---------
            |       |X: RFC 822|<------->|gateway|
            |       ------------         ---------
            | A           |                  ^
            \             |                  |
             \---------------------------------------------
                          |                  |
             /---------------------------------------------
            /             |                  |
            | B           |                  v
            |             |              -----------
            |             |              |Z: X.400 |
            |             |              -----------
            |             |                  .
            |             |                  .
            |             |                  .
            |             |                  .
            |             |                  .
            |             v                  v
            |        ------------         ---------
            |        |Y: RFC 822|<........|gateway|
            |        ------------         ---------

                    Fig. 3.1 The third party problem


         Now if Z wants to 'group reply' to both X and Y, his reply to Y
         will be routed over the gateway in country A, even though Y is
         located in the same country:

                     From: C=B;...;S=Z
                     To:   DD.RFC-822=Y(a)B; C=A;....;O=GW ,
                           DD.RFC-822=X(a)A; C=A;....;O=GW

         The best way to travel for a message from Z to Y would of
         course have been over the gateway in country B:

                     From: C=B;...;S=Z
                     To:   DD.RFC-822=Y(a)B; C=B;....;O=GW ,
                           DD.RFC-822=X(a)A; C=A;....;O=GW

         The third party problem is caused by the fact that routing
         information is mapped into addresses.

         Ideally, the third party problem shouldn't exist. After all,
         address mapping affects addresses, and an address is not a
         route.... The reality is different however. For instance, very



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 24]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


         few X.400 products are capable to route messages on the
         contents of a DDA (actually, only RFC 1327 gateways will be
         able to interpret this type of DDA, and who says that the reply
         will pass a local gateway on its route back?).  Similar
         limitations hold for the other direction: an RFC 822 based
         mailer is not even allowed (see [5]) to make routing decisions
         of the content of a left-hand-side encoded X.400 address if the
         domain part is not its own.  So in practice, addressing and
         (thus also mapping) will very well affect routing.

   To make mapping between addresses more user friendly, and to avoid
   the problems shown above, RFC 1327 allows for overruling the default
   left-hand-side encoding and DDA mapping algorithms. This is done by
   specifying associations (mapping rules) between certain domainparts
   and X.400 domains. An X.400 domain (for our purposes; CCITT has a
   narrower definition...) consists of the domain-related SAs of a
   Mnemonic O/R address (i.e., all SAs except PN and CN). The idea is to
   use the similarities between both address spaces, and directly map
   similar address parts onto each other. If, for the domain in the
   address to be mapped, an explicit mapping rule can be found, the
   mapping is performed between:

        localpart     <->   PersonalName
        domainpart    <->   X.400 domain

   The address information of the gateway is only used as an input
   parameter if no mapping rule can be found, i.e., if the address
   mapping must fall back to its default algorithm.

   The complete mapping function can thus be visualised as follows:


          address information of the gateway performing the mapping
                                      |
                                      v
                             +-----------------+
        RFC 822 address <--->| address mapping | <---> X.400 address
                             +-----------------+
                                      ^
                                      |
                    domain associations (mapping rules)

3.3.2.1. PersonalName and localpart mapping

   Since the mapping between these address parts is independent of the
   mapping rules that are used, and because it follows a simple, two-
   way algorithmic approach, this subject is discussed in a separate
   sub-chapter first.



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 25]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   The X.400 PersonalName consists of givenName, initials, and surName.
   RFC 1327 assumes that generationQualifier is not used.

   To map a localpart to an X.400 PN, the localpart is scanned for dots,
   which are considered delimiters between the components of PN, and
   also between single initials. In order not to put too much detail in
   this tutorial, only a few examples are shown here. For the detailed
   algorithm, see RFC 1327, chapter 4.2.1.

        Marshall.Rose             <->   G=Marshall;S=Rose
        M.T.Rose                  <->   I=MT;S=Rose
        Marshall.M.T.Rose         <->   G=Marshall;I=MT;S=Rose

   To map an X.400 PN to an RFC 822 localpart, take the non-empty PN
   attributes, put them into their hierarchical order (G I* S), and
   connect them with periods.

   Some exceptions are caused by the fact that left-hand-side encoding
   can also be mixed with exception mapping. This is shown in more
   detail in the following sub-chapters.

3.3.2.2. X.400 domain and domainpart mapping

   A mapping rule associates two domains: an X.400 domain and an RFC 822
   domain. The X.400 domain is written in the RFC 1327 domain notation
   (See 3.1.3.), so that both domains have the same hierarchical order.
   The domains are written on one line, separated by a '#' sign. For
   instance:

        arcom.ch#ADMD$arcom.C$ch#
        PRMD$tlec.ADMD$ade.C$nl#tlec.nl#

   A mapping rule must at least contain a top level domain and a country
   code. If an address must be mapped, a mapping rule with the longest
   domain match is sought. The associated domain in the mapping rule is
   used as the domain of the mapped address. The remaining domains are
   mapped one by one following the natural hierarchy. Concrete examples
   are shown in the following subchapters.

3.3.2.2.1. X.400 -> RFC 822

   As an example, assume the following mapping rule is defined:

           PRMD$tlec.ADMD$ade.C$nl#tlec.nl#







RARE Working Group on Mail and Messaging (WG-MSG)              [Page 26]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   Then the address C=nl; ADMD=ade; PRMD=tlec; O=you; OU=owe; S=plork

           S      OU  O  PRMD  ADMD  Country
           |      |   |  |     |     |
           plork owe you tlec  ade   nl

   would be mapped as follows. The Surname 'plork' is mapped to the
   localpart 'plork', see chapter 3.3.2.1. The domain

           localpart
              |  sdom3
              |    | sdom2
              |    |   |  sdom1
              |    |   |   |  top-level-domain
              |    |   |   |   |
           plork@         tlec.nl

   The remaining SAs (O and one OU) are mapped one by one following the
   natural hierarchy: O is mapped to sdom2, OU is mapped to sdom3:

           localpart
              | sdom3
              |  | sdom2
              |  |   |  sdom1
              |  |   |   |  top-level-domain
              |  |   |   |    |
           plork@owe.you.tlec.nl

   Thus the mapped address is:

           plork@owe.you.tlec.nl

   The table containing the listing of all such mapping rules, which is
   distributed to all gateways world-wide, is normally referred to as
   'mapping table 1'. Other commonly used filenames (also depending on
   which software your are using) are:

           'or2rfc'
           'mapping 1'
           'map1'
           'table 1'
           'X2R'

   As already announced, there is an exceptional case were localpart and
   PN are not directly mapped onto each other: sometimes it is necessary
   to use the localpart for other purposes. If the X.400 address
   contains attributes that would not allow for the simple mapping:




RARE Working Group on Mail and Messaging (WG-MSG)              [Page 27]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


           localpart     <->   PersonalName
           domainpart    <->   X.400 domain

   (e.g., spaces are not allowed in an RFC 822 domain, GQ and CN cannot
   be directly mapped into localpart, DDAs of another type than RFC-
   822), such attributes, together with the PN, are left-hand-side
   encoded. The domainpart must still be mapped according to the mapping
   rule as far as possible. This probably needs some examples:

           C=nl; ADMD=ade; PRMD=tlec; O=owe; OU=you; S=plork; GQ=jr
           ->
           /S=plork/GQ=jr/@you.owe.tlec.nl

           C=nl; ADMD=ade; PRMD=tlec; O=owe; OU=spc ctr; OU=u; S=plork
           ->
           "/S=plork/OU=u/OU=spc ctr/"@owe.tlec.nl

   Note that in the second example, 'O=owe' is still mapped to a
   subdomain following the natural hierarchy. The problems start with
   the space in 'OU=spc ctr'.

3.3.2.2.2. RFC 822 -> X.400

   As an example, assume the following mapping rule is defined:

           tlec.nl#PRMD$tlec.ADMD$ade.C$nl#

   Then the address 'plork@owe.you.tlec.nl' :

           localpart
              |  sdom3
              |    | sdom2
              |    |   |  sdom1
              |    |   |   |  top-level-domain
              |    |   |   |   |
           plork@owe.you.tlec.nl

   would be mapped as follows.

   The localpart 'plork' is mapped to 'S=plork', see chapter 3.3.2.1.

   The domain 'tlec.nl' is mapped according to the mapping rule:

           S     OU  OU  O  PRMD  ADMD  Country
           |                |     |    |
           plork            tlec  ade  nl

   The remaining domains (owe.you) are mapped one by one following the



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 28]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   natural hierarchy: sdom2 is mapped to O, sdom3 is mapped to OU:

           S     OU  OU  O  PRMD  ADMD  Country
           |         |   |  |     |     |
           plork     |   |  tlec  ade   nl
                     owe you

   Thus the mapped address is (in a readable notation):

           C=nl; ADMD=ade; PRMD=tlec; O=you; OU=owe; S=plork

   Had there been any left-hand-side encoded SAs in the localpart that
   didn't represent a complete mnemonic O/R address, the localpart would
   be mapped to those SAs. E.g.,

           "/S=plork/GQ=jr/OU=u/OU=spc ctr/"@owe.tlec.nl
           ->
           C=nl; ADMD=ade; PRMD=tlec; O=owe; OU=space ctr;
           OU=u; S=plork; GQ=jr

   This is necessary to reverse the special use of localpart to left-
   hand-side encode certain attributes. See 3.3.2.2.1.

   You might ask yourself by now why such rules are needed at all. Why
   don't we just use map1 in the other direction? The problem is that a
   symmetric mapping function (a bijection) would indeed be ideal, but
   it's not feasible. Asymmetric mappings exist for a number of reasons:

           - To make sure that uucp addresses etc. get routed over local
             gateways.

           - Preferring certain address forms, while still not forbidding
             others to use another form. Examples of such reasons are:

               - Phasing out old address forms.

               - If an RFC 822 address is mapped to ADMD= ; it means that
                 the X.400 mail can be routed over any ADMD in that
                 country. One single ADMD may of course send out an
                 address containing: ADMD=ade; . It must also be possible
                 to map such an address back.

   So we do need mapping rules from RFC 822 to X.400 too. The table
   containing the listing of all such mapping rules, which is
   distributed to all gateways world-wide, is normally referred to as on
   which software your are using) are:





RARE Working Group on Mail and Messaging (WG-MSG)              [Page 29]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


           'rfc2or'
           'mapping 2'
           'map2'
           'table 2'
           'R2X'

   If the RFC 822 localpart and/or domainpart contain characters that
   would not immediately fit in the value of a PN attribute (! % _), the
   mapping algorithm falls back to DDA mapping. In this case, the SAs
   that will be used are still determined by mapping the domainpart
   according to the mapping rule. In our case:

           100%user@work.tlec.nl
           ->
           DD.RFC-822=100(p)user(a)work.tlec.nl;
           C=nl; ADMD=ade; PRMD=tlec; O=work

   If no map2 rule can be found, a third table of rules is scanned: the
   gateway table. This table has the same syntax as mapping table 2, but
   its semantics are different. First of all, a domain that only has an
   entry in the gateway table is always mapped into an RFC 822 DDA. For
   a domain that is purely RFC 822 based, but whose mail may be relayed
   over an X.400 network, the gateway table associates with such a
   domain the SAs of the gateway to which the X.400 message should be
   routed. That gateway will then be responsible for gatewaying the
   message back into the RFC 822 world. E.g., if we have the gateway
   table entry:

           gov#PRMD$gateway.ADMD$Internet.C$us#

   (and we assume that no overruling map2 rule for the top level domain
   'gov' exists), this would force all gateways to perform the following
   mapping:

           bush@dole.gov
           ->
           DD.RFC-822=bush(a)dole.gov;
           C=us; ADMD=Internet; PRMD=gateway

   This is very similar to the default DDA mapping, except the SAs are
   those of a gateway that has declared to be responsible for a certain
   RFC 822 domain, not those of the local gateway. And thus, this
   mechanism helps avoid the third party problem discussed in chapter
   3.2.2.

   The table containing the listing of all such gateway rules, which is
   distributed to all gateways world-wide, is normally referred to as
   the 'gateway table'. Other commonly used filenames (also depending on



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 30]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   which software your are using) are:

           'rfc1148gate' {From the predecessor of RFC 1327, RFC 1148}
           'gate table'
           'GW'

   Only when no rule at all (map2 or gateway rule) is defined for a
   domain, the algorithm falls back to the default DDA mapping as
   described in 3.3.1.2.

3.4. Table co-ordination

   As already stated, the use of mapping tables will only function
   smoothly if all gateways in the world use the same tables. On the
   global level, the collection and distribution of RFC 1327 address
   mapping tables is co-ordinated by the MHS Co-ordination Service:

          SWITCH Head Office
          MHS Co-ordination Service
          Limmatquai 138
          CH-8001 Zurich, Europe
          Tel. +41 1 268 1550
          Fax. +41 1 268 1568

          RFC 822: project-team@switch.ch
          X.400:   C=ch;ADMD=arcom;PRMD=switch;O=switch;S=project-team;

   The procedures for collection and distribution of mapping rules can
   be found on the MHS Co-ordination Server, in the directory
   "/procedures".  Appendix D describes how this server can be accessed.

   If you want to define mapping rules for your own local domain, you
   can find the right contact person in your country or network (the
   gateway manager) on the same server, in the directory "/mhs-
   services".

3.5. Local additions

   Since certain networks want to define rules that should only be used
   within their networks, such rules should not be distributed world-
   wide. Consider two networks that both want to reach the old top-
   level-domain 'arpa' over their local gateway. They would both like to
   use a mapping 2 rule for this purpose:

           TLec in NL:     arpa#PRMD$gateway.ADMD$tlec.C$nl#

           SWITCH in CH:   arpa#PRMD$gateway.ADMD$switch.C$ch#




RARE Working Group on Mail and Messaging (WG-MSG)              [Page 31]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   (You may have noticed correctly that they should have defined such
   rules in the gateway table, but for the sake of the example, we
   assume they defined it in mapping table 2. This was the way things
   were done in the days of RFC 987, and many networks are still doing
   it this way these days.)

   Since a mapping table cannot contain two mapping rules with the same
   domain on the left hand side, such 'local mappings' are not
   distributed globally. There exists a RARE draft proposal [13] which
   defines a mechanism for allowing and automatically dealing with
   conflicting mapping rules, but this mechanism has not been
   implemented as to date. After having received the global mapping
   tables from the MHS Co-ordination Service, many networks add 'local'
   rules to map2 and the gateway table before installing them on their
   gateways. Note that the reverse mapping 2 rules for such local
   mappings _are_ globally unique, and can thus be distributed world-
   wide. This is even necessary, because addresses that were mapped with
   a local mapping rule may leak out to other networks (here comes the
   third party problem again...). Such other networks should at least be
   given the possibility to map the addresses back. So the global
   mapping table 1 would in this case contain the two rules:

           PRMD$gateway.ADMD$tlec.C$nl#arpa#
           PRMD$gateway.ADMD$switch.C$ch#arpa#

   Note that if such rules would have been defined as local gate table
   entries instead of map2 entries, there would have been no need to
   distribute the reverse mappings world-wide (the reverse mapping of a
   DDA encoded RFC 822 address is simply done by stripping the SAs, see
   3.3.1.1.).

3.6. Product specific formats

   Not all software uses the RFC 1327 format of the mapping tables
   internally. Almost all formats allow comments on a line starting with
   a # sign. Some examples of different formats:















RARE Working Group on Mail and Messaging (WG-MSG)              [Page 32]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


    RFC 1327

        # This is pure RFC 1327 format
        # table 1: X.400 -> RFC 822
        #
        PRMD$tlec.ADMD$ade.C$nl#tlec.nl#
        # etc.

        # table 2: RFC 822 -> X.400
        #
        arcom.ch#ADMD$arcom.C$ch#
        # etc.

    EAN

        # This is EAN format
        # It uses the readable format for X.400 domains and TABs
        # to make a 'readable mapping table format'.
        # table 1: X.400 -> RFC 822
        #
        P=tlec; A=ade; C=nl;       # tlec.nl
        # etc.

        # table 2: RFC 822 -> X.400
        #
        arcom.ch                   # A=arcom; C=ch;
        # etc.

    PP

        # This is PP format
        # table 1: X.400 -> RFC 822
        #
        PRMD$tlec.ADMD$ade.C$nl:tlec.nl
        # etc.

        # table 2: RFC 822 -> X.400
        #
        arcom.ch:ADMD$arcom.C$ch
        # etc.

   Most R&D networks have tools to automatically generate these formats
   from the original RFC 1327 tables;, some even distribute the tables
   within their networks in several formats. If you need mapping tables
   in a specific format, please contact your national or R&D network's
   gateway manager. See chapter 3.4.





RARE Working Group on Mail and Messaging (WG-MSG)              [Page 33]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


3.7. Guidelines for mapping rule definition

   Beware that defining mapping rules without knowing what you are doing
   can be disastrous not only for your network, but also for others. You
   should be rather safe if you follow at least these rules:

           - First of all, read this tutorial;.

           - Avoid local mappings; prefer gate table entries. (See chapter
             3.5)

           - Make sure any domain you map to can also be mapped back;.

           - Aim for symmetry.

           - Don't define a gateway table entry if the same domain already
             has a map2 entry. Such a rule would be redundant.

           - Map to "ADMD=0;" if you will not be connected to any ADMD for
             the time being.

           - Only map to "ADMD= ;" if you are indeed reachable through
             _any_ ADMD in your country.

           - Mind the difference between "PRMD=;" and "PRMD=@;" and make
             sure which one you need. (Try to avoid empty or unused
             attributes in the O/R address hierarchy from the beginning!)

           - Don't define mappings for domains over which you have no
             naming authority.

           - Before defining a mapping rule, make sure you have the
             permission from the naming authority of the domain you want
             to map to. Normally, this should be the same organisation as
             the mapping authority of the domain in the left hand side of
             the mapping rule. This principle is called 'administrative
             equivalence'.

           - Avoid redundant mappings. E.g., if all domains under 'tlec.nl'
             are in your control, don't define:

               first.tlec.nl#O$first.PRMD$tlec.ADMD$ade.C$nl#
               last.tlec.nl#O$last.PRMD$tlec.ADMD$ade.C$nl#
               always.tlec.nl#O$always.PRMD$tlec.ADMD$ade.C$nl#

             but rather have only one mapping rule:

               tlec.nl#PRMD$tlec.ADMD$ade.C$nl#



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 34]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993



           - Before introducing a new mapped version of a domain, make
             sure the world can route to that mapped domain;.

             E.g., If you are operating a PRMD: C=zz; ADMD=ade; PRMD=ergo;
             and you want to define the mapping rules:

               map1: PRMD$ergo.ADMD$ade.C$zz#ergo.zz#
               map2: ergo.zz#PRMD$ergo.ADMD$ade.C$zz#

             Make sure that ergo.zz (or at least all of its subdomains) is
             DNS routeable (register an MX or A record) and will be routed
             to a gateway that agreed to route the messages from the
             Internet to you over X.400.

             In the other direction, if you are operating the Internet
             domain cs.woodstock.edu, and you want to define a mapping for
             that domain:

               map2: cs.woodstock.edu#O$cs.PRMD$woodstock.ADMD$ .C$us#
               map1: O$cs.PRMD$woodstock.ADMD$ .C$us#cs.woodstock.edu#

             Make sure that C=us; ADMD= ; PRMD=woodstock; O=cs; (or at
             least all of its subdomains) is routeable in the X.400 world,
             and will be routed to a gateway that agreed to route the
             messages from X.400 to your RFC 822 domain over SMTP. Within
             the GO-MHS community, this would be done by registering a
             line in a so-called domain document, which will state to
             which mail relay this domain should be routed.

             Co-ordinate any such actions with your national or MHS'
             gateway manager. See chapter 3.4.

4. Conclusion

   Mail gatewaying remains a complicated subject. If after reading this
   tutorial, you feel you understand the basics, try solving some real-
   life problems. This is indeed a very rewarding area to work in: even
   after having worked with it for many years, you can make amazing
   discoveries every other week........











RARE Working Group on Mail and Messaging (WG-MSG)              [Page 35]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


Appendix A. References

   [1]  Postel, J., "Simple Mail Transfer Protocol", STD 10, RFC 821,
        USC/Information Sciences Institute, August 1982.

   [2]  Crocker, D., "Standard for the Format of ARPA Internet Text
        Messages", STD 11, RFC 822, University of Delaware, August 1982.

   [3]  Mockapetris, P., "Domain Names - Concepts and Facilities", and
        "Domain Names - Implementation and Specification", STD 13, RFCs
        1034 and 1035, USC/Information Sciences Institute, November
        1987.

   [4]  Kille, S., "Mapping Between X.400 and RFC 822", RFC 987, UK
        Academic Community Report (MG.19), UCL, June 1986.

   [5]  Braden, R., Editor, "Requirements for Internet Hosts --
        Application and Support", STD 3, RFC 1123, USC/Information
        Sciences Institute, October 1989.

   [6]  Postel, J., Editor, "Internet Official Protocol Standards", STD
        1, RFC 1500, USC/Information Sciences Institute, August 1993.

   [7]  Chapin, L., Chair, "The Internet Standards Process", RFC 1310,
        Internet Activities Board, March 1992.

   [8]  Kille, S., "Mapping between X.400(1988) / ISO 10021 and RFC
        822", RFC 1327 / RARE RTR 2, University College London, May
        1992.

   [9]  Kille, S., "X.400 1988 to 1984 downgrading", RFC 1328 / RARE RTR
        3, University College London, May 1992.

   [10] Plattner, B., and H. Lubich, "Electronic Mail Systems and
        Protocols Overview and Case Study", Proceedings of the IFIP WG
        6.5 International working conference on message handling systems
        and distributed applications; Costa Mesa 1988; North-Holland,
        1989.

   [11] Houttuin, J., "@route:100%name@address, a practical guide to MHS
        configuration", Top-Level EC, 1993, (not yet published).

   [12] Alvestrand, H., "Frequently asked questions on X.400", regularly
        posted on USEnet in newsgroup comp.protocols.iso.x400.

   [13] Houttuin, J., Hansen, K., and S. Aumont, "RFC 1327 Address
        Mapping Authorities", RARE WG-MSG Working Draft, Work in
        Progress, May 1993.



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 36]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


   [14] "COSINE MHS Pocket User Guide", COSINE MHS Project Team 1992.
        Also available in several languages from the MHS Co-ordination
        Server:/user-guides. See Appendix D.

   [15] Grimm, R., and S. Haug, "A Minimum Profile for RFC 987", GMD,
        November 1987; RARE MHS Project Team; July 1990. Also available
        from the MHS Co-ordination Server:/procedures/min-rfc987-
        profile. See Appendix D.

   [16] CCITT Recommendations X.400 - X.430. Data Communication
        Networks: Message Handling Systems.  CCITT Red Book, Vol. VIII -
        Fasc. VIII.7, Malaga-Torremolinos 1984.

   [17] CCITT Recommendations X.400 - X.420. Data Communication
        Networks: Message Handling Systems.  CCITT Blue Book, Vol. VIII
        - Fasc. VIII.7, Melbourne 1988.

Appendix B. Index

   <>

Appendix C. Abbreviations


      ADMD     Administration Management Domain
      ARPA     Advanced Research Projects Agency
      ASCII    American Standard Code for Information Exchange
      ASN.1    Abstract Syntax Notation One
      BCD      Binary-Coded Decimal
      BITNET   Because It's Time NETwork
      CCITT    Comite Consultatif International de Telegraphique et
               Telephonique
      COSINE   Co-operation for OSI networking in Europe
      DFN      Deutsches Forschungsnetz
      DL       Distribution List
      DNS      Domain Name System
      DoD      Department of Defense
      EBCDIC   Extended BCD Interchange Code
      IAB      Internet Architecture Board
      IEC      International Electrotechnical Commission
      IESG     Internet Engineering Steering Group
      IETF     Internet Engineering Task Force
      IP       Internet Protocol
      IPM      Inter-Personal Message
      IPMS     Inter-Personal Messaging Service
      IPN      Inter-Personal Notification
      ISO      International Organisation for Standardisation
      ISOC     Internet Society



RARE Working Group on Mail and Messaging (WG-MSG)              [Page 37]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


      ISODE    ISO Development Environment
      JNT      Joint Network Team (UK)
      JTC      Joint Technical Committee (ISO/IEC)
      MHS      Message Handling System
      MOTIS    Message-Oriented Text Interchange Systems
      MTA      Message Transfer Agent
      MTL      Message Transfer Layer
      MTS      Message Transfer System
      MX       Mail eXchanger
      OSI      Open Systems Interconnection
      OU(s)    Organizational Unit(s)
      PP       Mail gatewaying software (not an abbreviation)
      PRMD     Private Management Domain
      RARE     Reseaux Associes pour la Recherche Europeenne
      RFC      Request for comments
      RTC      RARE Technical Committee
      RTR      RARE Technical Report
      SMTP     simple mail transfer protocol
      STD      Internet Standard
      TCP      Transmission Control Protocol
      UUCP     Unix to Unix CoPy

Appendix D. How to access the MHS Co-ordination Server

   Here is an at-a-glance sheet on the access possibilities of the MHS
   Co-ordination server:

      E-mail

        address:

          RFC822: mhs-server@nic.switch.ch
          X.400:  S=mhs-server; OU1=nic; O=switch; P=switch; A=arcom;
                  C=CH

        body

          help                       # you receive this document
          index ['directory']        # you receive a directory listing
          send 'directory''filename' # you receive the specified file

      FTP

        address:  Internet: nic.switch.ch
        account:  cosine
        password: 'your email address'





RARE Working Group on Mail and Messaging (WG-MSG)              [Page 38]

RFC 1506        X.400-Internet Mail Gatewaying Tutorial      August 1993


      Interactive

        address:   Internet: nic.switch.ch
        address:   PSPDN:    +22847971014540
        address:   EMPB/IXI: 20432840100540
        account:   info
        directory: e-mail/COSINE-MHS/

      FTAM

        address:  Internet: nic.switch.ch
        address:  PSPDN   : +22847971014540
        address:  EMPB/IXI: 20432840100540
        address:  ISO CLNS: NSAP=39756f11112222223333aa0004000ae100,
                            TSEL=0103Hex
        account:  ANON

      gopher

        address:  Internet: nic.switch.ch

Security Considerations

   Security issues are not discussed in this memo.

Author's Address

   Jeroen Houttuin
   RARE Secretariat
   Singel 466-468
   NL-1017 AW Amsterdam
   Europe

   Tel. +31 20 6391131
   Fax. +31 20 6393289
   RFC 822: houttuin@rare.nl
   X.400:   C=nl;ADMD=400net;PRMD=surf;O=rare;S=houttuin














RARE Working Group on Mail and Messaging (WG-MSG)              [Page 39]