AMQP-0-9-1 XML

This is the final editorial draft 5 of AMQP0-91.xml. Please comment on this page, or by email to all participants of the 0-9-1 SIG.

The current editorial draft of the AMQP/0-9-1 specifications are here.

<?xml version = "1.0"?>
<!-- Editorial draft 5 -->
<!--
    Copyright Notice
    ================
    Copyright (c) 2006-2008 Cisco Systems, Credit Suisse, Deutsche Börse
    Systems, Envoy Technologies, Inc., Goldman Sachs, IONA Technologies PLC,
    iMatix Corporation, JPMorgan Chase Bank Inc. N.A, Novell, Rabbit
    Technologies Ltd., Red Hat, Inc., TWIST Process Innovations Ltd, WS02
    Inc. and 29West Inc. All rights reserved.

    License
    =======
    Cisco Systems, Credit Suisse, Deutsche Börse Systems, Envoy Technologies,
    Inc., Goldman Sachs, IONA Technologies PLC, iMatix Corporation, JPMorgan
    Chase Bank Inc. N.A, Novell, Rabbit Technologies Ltd., Red Hat, Inc.,
    TWIST Process Innovations Ltd, WS02, Inc. and 29West Inc. (collectively,
    the "Authors") each hereby grants to you a worldwide, perpetual,
    royalty-free, nontransferable, nonexclusive license to (i) copy, display,
    distribute and implement the Advanced Messaging Queue Protocol ("AMQP")
    Specification and (ii) the Licensed Claims that are held by the Authors,
    all for the purpose of implementing the Advanced Messaging Queue Protocol
    Specification. Your license and any rights under this Agreement will
    terminate immediately without notice from any Author if you bring any
    claim, suit, demand, or action related to the Advanced Messaging Queue
    Protocol Specification against any Author. Upon termination, you shall
    destroy all copies of the Advanced Messaging Queue Protocol Specification
    in your possession or control.

    As used hereunder, "Licensed Claims" means those claims of a patent or
    patent application, throughout the world, excluding design patents and
    design registrations, owned or controlled, or that can be sublicensed
    without fee and in compliance with the requirements of this Agreement,
    by an Author or its affiliates now or at any future time and which would
    necessarily be infringed by implementation of the Advanced Messaging
    Queue Protocol Specification. A claim is necessarily infringed hereunder
    only when it is not possible to avoid infringing it because there is no
    plausible non-infringing alternative for implementing the required
    portions of the Advanced Messaging Queue Protocol Specification.
    Notwithstanding the foregoing, Licensed Claims shall not include any
    claims other than as set forth above even if contained in the same patent
    as Licensed Claims; or that read solely on any implementations of any
    portion of the Advanced Messaging Queue Protocol Specification that are
    not required by the Advanced Messaging Queue ProtocolSpecification, or
    that, if licensed, would require a payment of royalties by the licensor
    to unaffiliated third parties. Moreover, Licensed Claims shall not
    include (i) any enabling technologies that may be necessary to make or
    use any Licensed Product but are not themselves expressly set forth in
    the Advanced Messaging Queue Protocol Specification (e.g., semiconductor
    manufacturing technology, compiler technology, object oriented
    technology, networking technology, operating system technology, and the
    like); or (ii) the implementation of other published standards developed
    elsewhere and merely referred to in the body of the Advanced Messaging
    Queue Protocol Specification, or (iii) any Licensed Product and any
    combinations thereof the purpose or function of which is not required
    for compliance with the Advanced Messaging Queue Protocol Specification.
    For purposes of this definition, the Advanced Messaging Queue Protocol
    Specification shall be deemed to include both architectural and
    interconnection requirements essential for interoperability and may also
    include supporting source code artifacts where such architectural,
    interconnection requirements and source code artifacts are expressly
    identified as being required or documentation to achieve compliance with
    the Advanced Messaging Queue Protocol Specification.

    As used hereunder, "Licensed Products" means only those specific portions
    of products (hardware, software or combinations thereof) that implement
    and are compliant with all relevant portions of the Advanced Messaging
    Queue Protocol Specification.

    The following disclaimers, which you hereby also acknowledge as to any
    use you may make of the Advanced Messaging Queue Protocol Specification:

    THE ADVANCED MESSAGING QUEUE PROTOCOL SPECIFICATION IS PROVIDED "AS IS,"
    AND THE AUTHORS MAKE NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
    IMPLIED, INCLUDING, BUT NOT LIMITED TO, WARRANTIES OF MERCHANTABILITY,
    FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, OR TITLE; THAT THE
    CONTENTS OF THE ADVANCED MESSAGING QUEUE PROTOCOL SPECIFICATION ARE
    SUITABLE FOR ANY PURPOSE; NOR THAT THE IMPLEMENTATION OF THE ADVANCED
    MESSAGING QUEUE PROTOCOL SPECIFICATION WILL NOT INFRINGE ANY THIRD PARTY
    PATENTS, COPYRIGHTS, TRADEMARKS OR OTHER RIGHTS.

    THE AUTHORS WILL NOT BE LIABLE FOR ANY DIRECT, INDIRECT, SPECIAL,
    INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF OR RELATING TO ANY
    USE, IMPLEMENTATION OR DISTRIBUTION OF THE ADVANCED MESSAGING QUEUE
    PROTOCOL SPECIFICATION.

    The name and trademarks of the Authors may NOT be used in any manner,
    including advertising or publicity pertaining to the Advanced Messaging
    Queue Protocol Specification or its contents without specific, written
    prior permission. Title to copyright in the Advanced Messaging Queue
    Protocol Specification will at all times remain with the Authors.

    No other rights are granted by implication, estoppel or otherwise.

    Upon termination of your license or rights under this Agreement, you
    shall destroy all copies of the Advanced Messaging Queue Protocol
    Specification in your possession or control.

    Trademarks
    ==========
    JPMorgan, JPMorgan Chase, Chase, the JPMorgan Chase logo and the
    Octagon Symbol are trademarks of JPMorgan Chase & Co.

    IMATIX and the iMatix logo are trademarks of iMatix Corporation sprl.

    IONA, IONA Technologies, and the IONA logos are trademarks of IONA
    Technologies PLC and/or its subsidiaries.

    LINUX is a trademark of Linus Torvalds. RED HAT and JBOSS are registered
    trademarks of Red Hat, Inc. in the US and other countries.

    Java, all Java-based trademarks and OpenOffice.org are trademarks of
    Sun Microsystems, Inc. in the United States, other countries, or both.

    Other company, product, or service names may be trademarks or service
    marks of others.

    Links to full AMQP specification:
    =================================
    http://www.amqp.org
-->

<!--
    <!DOCTYPE amqp SYSTEM "amqp.dtd">
-->

<!-- XML Notes

    We use entities to indicate repetition; attributes to indicate properties.

    We use the 'name' attribute as an identifier, usually within the context
    of the surrounding entities.

    We use spaces to seperate words in names, so that we can print names in
    their natural form depending on the context - underlines for source code,
    hyphens for written text, etc.

    We do not enforce any particular validation mechanism but we support all
    mechanisms.  The protocol definition conforms to a formal grammar that is
    published seperately in several technologies.

 -->

<amqp major = "0" minor = "9" revision = "1"
    port = "5672" comment = "AMQ Protocol version 0-9-1">
  <!--
      ======================================================
      ==       CONSTANTS
      ======================================================
  -->
  <!-- Frame types -->
  <constant name = "frame-method"     value = "1" />
  <constant name = "frame-header"     value = "2" />
  <constant name = "frame-body"       value = "3" />
  <constant name = "frame-heartbeat"  value = "8" />

  <!-- Protocol constants -->
  <constant name = "frame-min-size"   value = "4096" />
  <constant name = "frame-end"        value = "206" />

  <!-- Reply codes -->
  <constant name = "reply-success" value = "200">
    <doc>
      Indicates that the method completed successfully. This reply code is
      reserved for future use - the current protocol design does not use positive
      confirmation and reply codes are sent only in case of an error.
    </doc>
  </constant>

  <constant name = "content-too-large" value = "311" class = "soft-error">
    <doc>
      The client attempted to transfer content larger than the server could accept
      at the present time. The client may retry at a later time.
    </doc>
  </constant>

  <constant name = "no-consumers" value = "313" class = "soft-error">
    <doc>
      When the exchange cannot deliver to a consumer when the immediate flag is
      set. As a result of pending data on the queue or the absence of any
      consumers of the queue.
    </doc>
  </constant>

  <constant name = "connection-forced" value = "320" class = "hard-error">
    <doc>
      An operator intervened to close the connection for some reason. The client
      may retry at some later date.
    </doc>
  </constant>

  <constant name = "invalid-path" value = "402" class = "hard-error">
    <doc>
      The client tried to work with an unknown virtual host.
    </doc>
  </constant>

  <constant name = "access-refused" value = "403" class = "soft-error">
    <doc>
      The client attempted to work with a server entity to which it has no
      access due to security settings.
    </doc>
  </constant>

  <constant name = "not-found" value = "404" class = "soft-error">
    <doc>
      The client attempted to work with a server entity that does not exist.
    </doc>
  </constant>

  <constant name = "resource-locked" value = "405" class = "soft-error">
    <doc>
      The client attempted to work with a server entity to which it has no
      access because another client is working with it.
    </doc>
  </constant>

  <constant name = "precondition-failed" value = "406" class = "soft-error">
    <doc>
      The client requested a method that was not allowed because some precondition
      failed.
    </doc>
  </constant>

  <constant name = "frame-error" value = "501" class = "hard-error">
    <doc>
      The sender sent a malformed frame that the recipient could not decode.
      This strongly implies a programming error in the sending peer.
    </doc>
  </constant>

  <constant name = "syntax-error" value = "502" class = "hard-error">
    <doc>
      The sender sent a frame that contained illegal values for one or more
      fields. This strongly implies a programming error in the sending peer.
    </doc>
  </constant>

  <constant name = "command-invalid" value = "503" class = "hard-error">
    <doc>
      The client sent an invalid sequence of frames, attempting to perform an
      operation that was considered invalid by the server. This usually implies
      a programming error in the client.
    </doc>
  </constant>

  <constant name = "channel-error" value = "504" class = "hard-error">
    <doc>
      The client attempted to work with a channel that had not been correctly
      opened. This most likely indicates a fault in the client layer.
    </doc>
  </constant>

  <constant name = "unexpected-frame" value = "505" class = "hard-error">
    <doc>
      The peer sent a frame that was not expected, usually in the context of
      a content header and body.  This strongly indicates a fault in the peer's
      content processing.
    </doc>
  </constant>

  <constant name = "resource-error" value = "506" class = "hard-error">
    <doc>
      The server could not complete the method because it lacked sufficient
      resources. This may be due to the client creating too many of some type
      of entity.
    </doc>
  </constant>

  <constant name = "not-allowed" value = "530" class = "hard-error">
    <doc>
      The client tried to work with some entity in a manner that is prohibited
      by the server, due to security settings or by some other criteria.
    </doc>
  </constant>

  <constant name = "not-implemented" value = "540" class = "hard-error">
    <doc>
      The client tried to use functionality that is not implemented in the
      server.
    </doc>
  </constant>

  <constant name = "internal-error" value = "541" class = "hard-error">
    <doc>
      The server could not complete the method because of an internal error.
      The server may require intervention by an operator in order to resume
      normal operations.
    </doc>
  </constant>

  <!--
      ======================================================
      ==       DOMAIN TYPES
      ======================================================
  -->

  <domain name = "class-id" type = "short" />

  <domain name = "consumer-tag" type = "shortstr" label = "consumer tag">
    <doc>
      Identifier for the consumer, valid within the current channel.
    </doc>
  </domain>

  <domain name = "delivery-tag" type = "longlong" label = "server-assigned delivery tag">
    <doc>
      The server-assigned and channel-specific delivery tag
    </doc>
    <rule name = "channel-local">
      <doc>
        The delivery tag is valid only within the channel from which the message was
        received. I.e. a client MUST NOT receive a message on one channel and then
        acknowledge it on another.
      </doc>
    </rule>
    <rule name = "non-zero">
      <doc>
        The server MUST NOT use a zero value for delivery tags. Zero is reserved
        for client use, meaning "all messages so far received".
      </doc>
    </rule>
  </domain>

  <domain name = "exchange-name" type = "shortstr" label = "exchange name">
    <doc>
      The exchange name is a client-selected string that identifies the exchange for
      publish methods.
    </doc>
    <assert check = "length" value = "127" />
    <assert check = "regexp" value = "^[a-zA-Z0-9-_.:]*$" />
  </domain>

  <domain name = "method-id" type = "short" />

  <domain name = "no-ack" type = "bit" label = "no acknowledgement needed">
    <doc>
      If this field is set the server does not expect acknowledgements for
      messages. That is, when a message is delivered to the client the server
      assumes the delivery will succeed and immediately dequeues it. This
      functionality may increase performance but at the cost of reliability.
      Messages can get lost if a client dies before they are delivered to the
      application.
    </doc>
  </domain>

  <domain name = "no-local" type = "bit" label = "do not deliver own messages">
    <doc>
      If the no-local field is set the server will not send messages to the connection that
      published them.
    </doc>
  </domain>

  <domain name = "no-wait" type = "bit" label = "do not send reply method">
    <doc>
      If set, the server will not respond to the method. The client should not wait
      for a reply method. If the server could not complete the method it will raise a
      channel or connection exception.
    </doc>
  </domain>

  <domain name = "path" type = "shortstr">
    <doc>
      Unconstrained.
    </doc>
    <assert check = "notnull" />
    <assert check = "length" value = "127" />
  </domain>

  <domain name = "peer-properties" type = "table">
    <doc>
      This table provides a set of peer properties, used for identification, debugging,
      and general information.
    </doc>
  </domain>

  <domain name = "queue-name" type = "shortstr" label = "queue name">
    <doc>
      The queue name identifies the queue within the vhost.  In methods where the queue
      name may be blank, and that has no specific significance, this refers to the
      'current' queue for the channel, meaning the last queue that the client declared
      on the channel.  If the client did not declare a queue, and the method needs a
      queue name, this will result in a 502 (syntax error) channel exception.
    </doc>
    <assert check = "length" value = "127" />
    <assert check = "regexp" value = "^[a-zA-Z0-9-_.:]*$" />
  </domain>

  <domain name = "redelivered" type = "bit" label = "message is being redelivered">
    <doc>
      This indicates that the message has been previously delivered to this or
      another client.
    </doc>
    <rule name = "implementation">
      <doc>
        The server SHOULD try to signal redelivered messages when it can. When
        redelivering a message that was not successfully acknowledged, the server
        SHOULD deliver it to the original client if possible.
      </doc>
      <doc type = "scenario">
        Declare a shared queue and publish a message to the queue.  Consume the
        message using explicit acknowledgements, but do not acknowledge the
        message.  Close the connection, reconnect, and consume from the queue
        again.  The message should arrive with the redelivered flag set.
      </doc>
    </rule>
    <rule name = "hinting">
      <doc>
        The client MUST NOT rely on the redelivered field but should take it as a
        hint that the message may already have been processed. A fully robust
        client must be able to track duplicate received messages on non-transacted,
        and locally-transacted channels.
      </doc>
    </rule>
  </domain>

  <domain name = "message-count" type = "long" label = "number of messages in queue">
    <doc>
      The number of messages in the queue, which will be zero for newly-declared
      queues. This is the number of messages present in the queue, and committed
      if the channel on which they were published is transacted, that are not
      waiting acknowledgement.
    </doc>
  </domain>

  <domain name = "reply-code" type = "short" label = "reply code from server">
    <doc>
      The reply code. The AMQ reply codes are defined as constants at the start
      of this formal specification.
    </doc>
    <assert check = "notnull" />
  </domain>

  <domain name = "reply-text" type = "shortstr" label = "localised reply text">
    <doc>
      The localised reply text. This text can be logged as an aid to resolving
      issues.
    </doc>
    <assert check = "notnull" />
  </domain>

  <!-- Elementary domains -->
  <domain name = "bit"        type = "bit"       label = "single bit" />
  <domain name = "octet"      type = "octet"     label = "single octet" />
  <domain name = "short"      type = "short"     label = "16-bit integer" />
  <domain name = "long"       type = "long"      label = "32-bit integer" />
  <domain name = "longlong"   type = "longlong"  label = "64-bit integer" />
  <domain name = "shortstr"   type = "shortstr"  label = "short string" />
  <domain name = "longstr"    type = "longstr"   label = "long string" />
  <domain name = "timestamp"  type = "timestamp" label = "64-bit timestamp" />
  <domain name = "table"      type = "table"     label = "field table" />

  <!-- ==  CONNECTION  ======================================================= -->

  <class name = "connection" handler = "connection" index = "10" label = "work with socket connections">
    <doc>
      The connection class provides methods for a client to establish a network connection to
      a server, and for both peers to operate the connection thereafter.
    </doc>

    <doc type = "grammar">
      connection          = open-connection *use-connection close-connection
      open-connection     = C:protocol-header
                            S:START C:START-OK
                            *challenge
                            S:TUNE C:TUNE-OK
                            C:OPEN S:OPEN-OK
      challenge           = S:SECURE C:SECURE-OK
      use-connection      = *channel
      close-connection    = C:CLOSE S:CLOSE-OK
                          / S:CLOSE C:CLOSE-OK
    </doc>

    <chassis name = "server" implement = "MUST" />
    <chassis name = "client" implement = "MUST" />

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "start" synchronous = "1" index = "10" label = "start connection negotiation">
      <doc>
        This method starts the connection negotiation process by telling the client the
        protocol version that the server proposes, along with a list of security mechanisms
        which the client can use for authentication.
      </doc>

      <rule name = "protocol-name">
        <doc>
          If the server cannot support the protocol specified in the protocol header,
          it MUST respond with a valid protocol header and then close the socket
          connection.
        </doc>
        <doc type = "scenario">
          The client sends a protocol header containing an invalid protocol name.
          The server MUST respond by sending a valid protocol header and then closing
          the connection.
        </doc>
      </rule>
      <rule name = "server-support">
        <doc>
          The server MUST provide a protocol version that is lower than or equal to
          that requested by the client in the protocol header.
        </doc>
        <doc type = "scenario">
          The client requests a protocol version that is higher than any valid
          implementation, e.g. 2.0.  The server must respond with a protocol header
          indicating its supported protocol version, e.g. 1.0.
        </doc>
      </rule>
      <rule name = "client-support">
        <doc>
          If the client cannot handle the protocol version suggested by the server
          it MUST close the socket connection without sending any further data.
        </doc>
        <doc type = "scenario">
          The server sends a protocol version that is lower than any valid
          implementation, e.g. 0.1.  The client must respond by closing the
          connection without sending any further data.
        </doc>
      </rule>

      <chassis name = "client" implement = "MUST" />
      <response name = "start-ok" />

      <field name = "version-major" domain = "octet" label = "protocol major version">
        <doc>
          The major version number can take any value from 0 to 99 as defined in the
          AMQP specification.
        </doc>
      </field>

      <field name = "version-minor" domain = "octet" label = "protocol minor version">
        <doc>
          The minor version number can take any value from 0 to 99 as defined in the
          AMQP specification.
        </doc>
      </field>

      <field name = "server-properties" domain = "peer-properties" label = "server properties">
        <rule name = "required-fields">
          <doc>
            The properties SHOULD contain at least these fields: "host", specifying the
            server host name or address, "product", giving the name of the server product,
            "version", giving the name of the server version, "platform", giving the name
            of the operating system, "copyright", if appropriate, and "information", giving
            other general information.
          </doc>
          <doc type = "scenario">
            Client connects to server and inspects the server properties. It checks for
            the presence of the required fields.
          </doc>
        </rule>
      </field>

      <field name = "mechanisms" domain = "longstr" label = "available security mechanisms">
        <doc>
          A list of the security mechanisms that the server supports, delimited by spaces.
        </doc>
        <assert check = "notnull" />
      </field>

      <field name = "locales" domain = "longstr" label = "available message locales">
        <doc>
          A list of the message locales that the server supports, delimited by spaces. The
          locale defines the language in which the server will send reply texts.
        </doc>
        <rule name = "required-support">
          <doc>
            The server MUST support at least the en_US locale.
          </doc>
          <doc type = "scenario">
            Client connects to server and inspects the locales field. It checks for
            the presence of the required locale(s).
          </doc>
        </rule>
        <assert check = "notnull" />
      </field>
    </method>

    <method name = "start-ok" synchronous = "1" index = "11"
      label = "select security mechanism and locale">
      <doc>
        This method selects a SASL security mechanism.
      </doc>

      <chassis name = "server" implement = "MUST" />

      <field name = "client-properties" domain = "peer-properties" label = "client properties">
        <rule name = "required-fields">
          <!-- This rule is not testable from the client side -->
          <doc>
            The properties SHOULD contain at least these fields: "product", giving the name
            of the client product, "version", giving the name of the client version, "platform",
            giving the name of the operating system, "copyright", if appropriate, and
            "information", giving other general information.
          </doc>
        </rule>
      </field>

      <field name = "mechanism" domain = "shortstr" label = "selected security mechanism">
        <doc>
          A single security mechanisms selected by the client, which must be one of those
          specified by the server.
        </doc>
        <rule name = "security">
          <doc>
            The client SHOULD authenticate using the highest-level security profile it
            can handle from the list provided by the server.
          </doc>
        </rule>
        <rule name = "validity">
          <doc>
            If the mechanism field does not contain one of the security mechanisms
            proposed by the server in the Start method, the server MUST close the
            connection without sending any further data.
          </doc>
          <doc type = "scenario">
            Client connects to server and sends an invalid security mechanism. The
            server must respond by closing the connection (a socket close, with no
            connection close negotiation).
          </doc>
        </rule>
        <assert check = "notnull" />
      </field>

      <field name = "response" domain = "longstr" label = "security response data">
        <doc>
          A block of opaque data passed to the security mechanism. The contents of this
          data are defined by the SASL security mechanism.
        </doc>
        <assert check = "notnull" />
      </field>

      <field name = "locale" domain = "shortstr" label = "selected message locale">
        <doc>
          A single message locale selected by the client, which must be one of those
          specified by the server.
        </doc>
        <assert check = "notnull" />
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "secure" synchronous = "1" index = "20" label = "security mechanism challenge">
      <doc>
        The SASL protocol works by exchanging challenges and responses until both peers have
        received sufficient information to authenticate each other. This method challenges
        the client to provide more information.
      </doc>

      <chassis name = "client" implement = "MUST" />
      <response name = "secure-ok" />

      <field name = "challenge" domain = "longstr" label = "security challenge data">
        <doc>
          Challenge information, a block of opaque binary data passed to the security
          mechanism.
        </doc>
      </field>
    </method>

    <method name = "secure-ok" synchronous = "1" index = "21" label = "security mechanism response">
      <doc>
        This method attempts to authenticate, passing a block of SASL data for the security
        mechanism at the server side.
      </doc>

      <chassis name = "server" implement = "MUST" />

      <field name = "response" domain = "longstr" label = "security response data">
        <doc>
          A block of opaque data passed to the security mechanism. The contents of this
          data are defined by the SASL security mechanism.
        </doc>
        <assert check = "notnull" />
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "tune" synchronous = "1" index = "30"
      label = "propose connection tuning parameters">
      <doc>
        This method proposes a set of connection configuration values to the client. The
        client can accept and/or adjust these.
      </doc>

      <chassis name = "client" implement = "MUST" />

      <response name = "tune-ok" />

      <field name = "channel-max" domain = "short" label = "proposed maximum channels">
        <doc>
          Specifies highest channel number that the server permits.  Usable channel numbers
          are in the range 1..channel-max.  Zero indicates no specified limit.
        </doc>
      </field>

      <field name = "frame-max" domain = "long" label = "proposed maximum frame size">
        <doc>
          The largest frame size that the server proposes for the connection, including
          frame header and end-byte.  The client can negotiate a lower value. Zero means
          that the server does not impose any specific limit but may reject very large
          frames if it cannot allocate resources for them.
        </doc>
        <rule name = "minimum">
          <doc>
            Until the frame-max has been negotiated, both peers MUST accept frames of up
            to frame-min-size octets large, and the minimum negotiated value for frame-max
            is also frame-min-size.
          </doc>
          <doc type = "scenario">
            Client connects to server and sends a large properties field, creating a frame
            of frame-min-size octets.  The server must accept this frame.
          </doc>
        </rule>
      </field>

      <field name = "heartbeat" domain = "short" label = "desired heartbeat delay">
        <doc>
          The delay, in seconds, of the connection heartbeat that the server wants.
          Zero means the server does not want a heartbeat.
        </doc>
      </field>
    </method>

    <method name = "tune-ok" synchronous = "1" index = "31"
      label = "negotiate connection tuning parameters">
      <doc>
        This method sends the client's connection tuning parameters to the server.
        Certain fields are negotiated, others provide capability information.
      </doc>

      <chassis name = "server" implement = "MUST" />

      <field name = "channel-max" domain = "short" label = "negotiated maximum channels">
        <doc>
          The maximum total number of channels that the client will use per connection.
        </doc>
        <rule name = "upper-limit">
          <doc>
            If the client specifies a channel max that is higher than the value provided
            by the server, the server MUST close the connection without attempting a
            negotiated close.  The server may report the error in some fashion to assist
            implementors.
          </doc>
        </rule>
        <assert check = "notnull" />
        <assert check = "le" method = "tune" field = "channel-max" />
      </field>

      <field name = "frame-max" domain = "long" label = "negotiated maximum frame size">
        <doc>
          The largest frame size that the client and server will use for the connection.
          Zero means that the client does not impose any specific limit but may reject
          very large frames if it cannot allocate resources for them. Note that the
          frame-max limit applies principally to content frames, where large contents can
          be broken into frames of arbitrary size.
        </doc>
        <rule name = "minimum">
          <doc>
            Until the frame-max has been negotiated, both peers MUST accept frames of up
            to frame-min-size octets large, and the minimum negotiated value for frame-max
            is also frame-min-size.
          </doc>
        </rule>
        <rule name = "upper-limit">
          <doc>
            If the client specifies a frame max that is higher than the value provided
            by the server, the server MUST close the connection without attempting a
            negotiated close. The server may report the error in some fashion to assist
            implementors.
          </doc>
        </rule>
      </field>

      <field name = "heartbeat" domain = "short" label = "desired heartbeat delay">
        <doc>
          The delay, in seconds, of the connection heartbeat that the client wants. Zero
          means the client does not want a heartbeat.
        </doc>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "open" synchronous = "1" index = "40" label = "open connection to virtual host">
      <doc>
        This method opens a connection to a virtual host, which is a collection of
        resources, and acts to separate multiple application domains within a server.
        The server may apply arbitrary limits per virtual host, such as the number
        of each type of entity that may be used, per connection and/or in total.
      </doc>

      <chassis name = "server" implement = "MUST" />
      <response name = "open-ok" />

      <field name = "virtual-host" domain = "path" label = "virtual host name">
        <doc>
          The name of the virtual host to work with.
        </doc>
        <rule name = "separation">
          <doc>
            If the server supports multiple virtual hosts, it MUST enforce a full
            separation of exchanges, queues, and all associated entities per virtual
            host. An application, connected to a specific virtual host, MUST NOT be able
            to access resources of another virtual host.
          </doc>
        </rule>
        <rule name = "security">
          <doc>
            The server SHOULD verify that the client has permission to access the
            specified virtual host.
          </doc>
        </rule>
      </field>
      <!-- Deprecated: "capabilities", must be zero -->
      <field name = "reserved-1" type = "shortstr" reserved = "1" />
      <!-- Deprecated: "insist", must be zero -->
      <field name = "reserved-2" type = "bit" reserved = "1" />
    </method>

    <method name = "open-ok" synchronous = "1" index = "41" label = "signal that connection is ready">
      <doc>
        This method signals to the client that the connection is ready for use.
      </doc>
      <chassis name = "client" implement = "MUST" />
      <!-- Deprecated: "known-hosts", must be zero -->
      <field name = "reserved-1" type = "shortstr" reserved = "1" />
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "close" synchronous = "1" index = "50" label = "request a connection close">
      <doc>
        This method indicates that the sender wants to close the connection. This may be
        due to internal conditions (e.g. a forced shut-down) or due to an error handling
        a specific method, i.e. an exception. When a close is due to an exception, the
        sender provides the class and method id of the method which caused the exception.
      </doc>
      <rule name = "stability">
        <doc>
          After sending this method, any received methods except Close and Close-OK MUST
          be discarded.  The response to receiving a Close after sending Close must be to
          send Close-Ok.
        </doc>
      </rule>

      <chassis name = "client" implement = "MUST" />
      <chassis name = "server" implement = "MUST" />
      <response name = "close-ok" />

      <field name = "reply-code" domain = "reply-code" />
      <field name = "reply-text" domain = "reply-text" />

      <field name = "class-id" domain = "class-id" label = "failing method class">
        <doc>
          When the close is provoked by a method exception, this is the class of the
          method.
        </doc>
      </field>

      <field name = "method-id" domain = "method-id" label = "failing method ID">
        <doc>
          When the close is provoked by a method exception, this is the ID of the method.
        </doc>
      </field>
    </method>

    <method name = "close-ok" synchronous = "1" index = "51" label = "confirm a connection close">
      <doc>
        This method confirms a Connection.Close method and tells the recipient that it is
        safe to release resources for the connection and close the socket.
      </doc>
      <rule name = "reporting">
        <doc>
          A peer that detects a socket closure without having received a Close-Ok
          handshake method SHOULD log the error.
        </doc>
      </rule>
      <chassis name = "client" implement = "MUST" />
      <chassis name = "server" implement = "MUST" />
    </method>
  </class>

  <!-- ==  CHANNEL  ========================================================== -->

  <class name = "channel" handler = "channel" index = "20" label = "work with channels">
    <doc>
      The channel class provides methods for a client to establish a channel to a
      server and for both peers to operate the channel thereafter.
    </doc>

    <doc type = "grammar">
      channel             = open-channel *use-channel close-channel
      open-channel        = C:OPEN S:OPEN-OK
      use-channel         = C:FLOW S:FLOW-OK
                          / S:FLOW C:FLOW-OK
                          / functional-class
      close-channel       = C:CLOSE S:CLOSE-OK
                          / S:CLOSE C:CLOSE-OK
    </doc>

    <chassis name = "server" implement = "MUST" />
    <chassis name = "client" implement = "MUST" />

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "open" synchronous = "1" index = "10" label = "open a channel for use">
      <doc>
        This method opens a channel to the server.
      </doc>
      <rule name = "state" on-failure = "channel-error">
        <doc>
          The client MUST NOT use this method on an already-opened channel.
        </doc>
        <doc type = "scenario">
          Client opens a channel and then reopens the same channel.
        </doc>
      </rule>
      <chassis name = "server" implement = "MUST" />
      <response name = "open-ok" />
      <!-- Deprecated: "out-of-band", must be zero -->
      <field name = "reserved-1" type = "shortstr" reserved = "1" />
    </method>

    <method name = "open-ok" synchronous = "1" index = "11" label = "signal that the channel is ready">
      <doc>
        This method signals to the client that the channel is ready for use.
      </doc>
      <chassis name = "client" implement = "MUST" />
      <!-- Deprecated: "channel-id", must be zero -->
      <field name = "reserved-1" type = "longstr" reserved = "1" />
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "flow" synchronous = "1" index = "20" label = "enable/disable flow from peer">
      <doc>
        This method asks the peer to pause or restart the flow of content data sent by
        a consumer. This is a simple flow-control mechanism that a peer can use to avoid
        overflowing its queues or otherwise finding itself receiving more messages than
        it can process. Note that this method is not intended for window control. It does
        not affect contents returned by Basic.Get-Ok methods.
      </doc>

      <rule name = "initial-state">
        <doc>
          When a new channel is opened, it is active (flow is active). Some applications
          assume that channels are inactive until started. To emulate this behaviour a
          client MAY open the channel, then pause it.
        </doc>
      </rule>

      <rule name = "bidirectional">
        <doc>
          When sending content frames, a peer SHOULD monitor the channel for incoming
          methods and respond to a Channel.Flow as rapidly as possible.
        </doc>
      </rule>

      <rule name = "throttling">
        <doc>
          A peer MAY use the Channel.Flow method to throttle incoming content data for
          internal reasons, for example, when exchanging data over a slower connection.
        </doc>
      </rule>

      <rule name = "expected-behaviour">
        <doc>
          The peer that requests a Channel.Flow method MAY disconnect and/or ban a peer
          that does not respect the request.  This is to prevent badly-behaved clients
          from overwhelming a server.
        </doc>
      </rule>

      <chassis name = "server" implement = "MUST" />
      <chassis name = "client" implement = "MUST" />

      <response name = "flow-ok" />

      <field name = "active" domain = "bit" label = "start/stop content frames">
        <doc>
          If 1, the peer starts sending content frames. If 0, the peer stops sending
          content frames.
        </doc>
      </field>
    </method>

    <method name = "flow-ok" index = "21" label = "confirm a flow method">
      <doc>
        Confirms to the peer that a flow command was received and processed.
      </doc>
      <chassis name = "server" implement = "MUST" />
      <chassis name = "client" implement = "MUST" />
      <field name = "active" domain = "bit" label = "current flow setting">
        <doc>
          Confirms the setting of the processed flow method: 1 means the peer will start
          sending or continue to send content frames; 0 means it will not.
        </doc>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "close" synchronous = "1" index = "40" label = "request a channel close">
      <doc>
        This method indicates that the sender wants to close the channel. This may be due to
        internal conditions (e.g. a forced shut-down) or due to an error handling a specific
        method, i.e. an exception. When a close is due to an exception, the sender provides
        the class and method id of the method which caused the exception.
      </doc>
      <rule name = "stability">
        <doc>
          After sending this method, any received methods except Close and Close-OK MUST
          be discarded.  The response to receiving a Close after sending Close must be to
          send Close-Ok.
        </doc>
      </rule>

      <chassis name = "client" implement = "MUST" />
      <chassis name = "server" implement = "MUST" />
      <response name = "close-ok" />

      <field name = "reply-code" domain = "reply-code" />
      <field name = "reply-text" domain = "reply-text" />

      <field name = "class-id" domain = "class-id" label = "failing method class">
        <doc>
          When the close is provoked by a method exception, this is the class of the
          method.
        </doc>
      </field>

      <field name = "method-id" domain = "method-id" label = "failing method ID">
        <doc>
          When the close is provoked by a method exception, this is the ID of the method.
        </doc>
      </field>
    </method>

    <method name = "close-ok" synchronous = "1" index = "41" label = "confirm a channel close">
      <doc>
        This method confirms a Channel.Close method and tells the recipient that it is safe
        to release resources for the channel.
      </doc>
      <rule name = "reporting">
        <doc>
          A peer that detects a socket closure without having received a Channel.Close-Ok
          handshake method SHOULD log the error.
        </doc>
      </rule>
      <chassis name = "client" implement = "MUST" />
      <chassis name = "server" implement = "MUST" />
    </method>
  </class>

  <!-- ==  EXCHANGE  ========================================================= -->

  <class name = "exchange" handler = "channel" index = "40" label = "work with exchanges">
    <doc>
      Exchanges match and distribute messages across queues. Exchanges can be configured in
      the server or declared at runtime.
    </doc>

    <doc type = "grammar">
      exchange            = C:DECLARE  S:DECLARE-OK
                          / C:DELETE   S:DELETE-OK
    </doc>

    <chassis name = "server" implement = "MUST" />
    <chassis name = "client" implement = "MUST" />

    <rule name = "required-types">
      <doc>
        The server MUST implement these standard exchange types: fanout, direct.
      </doc>
      <doc type = "scenario">
        Client attempts to declare an exchange with each of these standard types.
      </doc>
    </rule>
    <rule name = "recommended-types">
      <doc>
        The server SHOULD implement these standard exchange types: topic, headers.
      </doc>
      <doc type = "scenario">
        Client attempts to declare an exchange with each of these standard types.
      </doc>
    </rule>
    <rule name = "required-instances">
      <doc>
        The server MUST, in each virtual host, pre-declare an exchange instance
        for each standard exchange type that it implements, where the name of the
        exchange instance, if defined, is "amq." followed by the exchange type name.
      </doc>
      <doc>
        The server MUST, in each virtual host, pre-declare at least two direct
        exchange instances: one named "amq.direct", the other with no public name
        that serves as a default  exchange for Publish methods.
      </doc>
      <doc type = "scenario">
        Client declares a temporary queue and attempts to bind to each required
        exchange instance ("amq.fanout", "amq.direct", "amq.topic", and "amq.headers"
        if those types are defined).
      </doc>
    </rule>
    <rule name = "default-exchange">
      <doc>
        The server MUST pre-declare a direct exchange with no public name to act as
        the default exchange for content Publish methods and for default queue bindings.
      </doc>
      <doc type = "scenario">
        Client checks that the default exchange is active by specifying a queue
        binding with no exchange name, and publishing a message with a suitable
        routing key but without specifying the exchange name, then ensuring that
        the message arrives in the queue correctly.
      </doc>
    </rule>
    <rule name = "default-access">
      <doc>
        The server MUST NOT allow clients to access the default exchange except
        by specifying an empty exchange name in the Queue.Bind and content Publish
        methods.
      </doc>
    </rule>
    <rule name = "extensions">
      <doc>
        The server MAY implement other exchange types as wanted.
      </doc>
    </rule>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "declare" synchronous = "1" index = "10" label = "verify exchange exists, create if needed">
      <doc>
        This method creates an exchange if it does not already exist, and if the exchange
        exists, verifies that it is of the correct and expected class.
      </doc>
      <rule name = "minimum">
        <doc>
          The server SHOULD support a minimum of 16 exchanges per virtual host and
          ideally, impose no limit except as defined by available resources.
        </doc>
        <doc type = "scenario">
          The client declares as many exchanges as it can until the server reports
          an error; the number of exchanges successfully declared must be at least
          sixteen.
        </doc>
      </rule>

      <chassis name = "server" implement = "MUST" />
      <response name = "declare-ok" />

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "exchange" domain = "exchange-name">
        <rule name = "reserved" on-failure = "access-refused">
          <doc>
            Exchange names starting with "amq." are reserved for pre-declared and
            standardised exchanges. The client MAY declare an exchange starting with
            "amq." if the passive option is set, or the exchange already exists.
          </doc>
          <doc type = "scenario">
            The client attempts to declare a non-existing exchange starting with
            "amq." and with the passive option set to zero.
          </doc>
        </rule>
        <rule name = "syntax" on-failure = "precondition-failed">
          <doc>
            The exchange name consists of a non-empty sequence of these characters:
            letters, digits, hyphen, underscore, period, or colon.
          </doc>
          <doc type = "scenario">
            The client attempts to declare an exchange with an illegal name.
          </doc>
        </rule>
        <assert check = "notnull" />
      </field>

      <field name = "type" domain = "shortstr" label = "exchange type">
        <doc>
          Each exchange belongs to one of a set of exchange types implemented by the
          server. The exchange types define the functionality of the exchange - i.e. how
          messages are routed through it. It is not valid or meaningful to attempt to
          change the type of an existing exchange.
        </doc>
        <rule name = "typed" on-failure = "not-allowed">
          <doc>
            Exchanges cannot be redeclared with different types.  The client MUST not
            attempt to redeclare an existing exchange with a different type than used
            in the original Exchange.Declare method.
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
        <rule name = "support" on-failure = "command-invalid">
          <doc>
            The client MUST NOT attempt to declare an exchange with a type that the
            server does not support.
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
      </field>

      <field name = "passive" domain = "bit" label = "do not create exchange">
        <doc>
          If set, the server will reply with Declare-Ok if the exchange already
          exists with the same name, and raise an error if not.  The client can
          use this to check whether an exchange exists without modifying the
          server state. When set, all other method fields except name and no-wait
          are ignored.  A declare with both passive and no-wait has no effect.
          Arguments are compared for semantic equivalence.
        </doc>
        <rule name = "not-found">
          <doc>
            If set, and the exchange does not already exist, the server MUST
            raise a channel exception with reply code 404 (not found).
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
        <rule name = "equivalent">
          <doc>
            If not set and the exchange exists, the server MUST check that the
            existing exchange has the same values for type, durable, and arguments
            fields.  The server MUST respond with Declare-Ok if the requested
            exchange matches these fields, and MUST raise a channel exception if
            not.
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
      </field>

      <field name = "durable" domain = "bit" label = "request a durable exchange">
        <doc>
          If set when creating a new exchange, the exchange will be marked as durable.
          Durable exchanges remain active when a server restarts. Non-durable exchanges
          (transient exchanges) are purged if/when a server restarts.
        </doc>
        <rule name = "support">
          <doc>
            The server MUST support both durable and transient exchanges.
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
      </field>

      <!-- Deprecated: "auto-delete", must be zero -->
      <field name = "reserved-2" type = "bit" reserved = "1" />
      <!-- Deprecated: "internal", must be zero -->
      <field name = "reserved-3" type = "bit" reserved = "1" />
      <field name = "no-wait" domain = "no-wait" />

      <field name = "arguments" domain = "table" label = "arguments for declaration">
        <doc>
          A set of arguments for the declaration. The syntax and semantics of these
          arguments depends on the server implementation.
        </doc>
      </field>
    </method>

    <method name = "declare-ok" synchronous = "1" index = "11" label = "confirm exchange declaration">
      <doc>
        This method confirms a Declare method and confirms the name of the exchange,
        essential for automatically-named exchanges.
      </doc>
      <chassis name = "client" implement = "MUST" />
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "delete" synchronous = "1" index = "20" label = "delete an exchange">
      <doc>
        This method deletes an exchange. When an exchange is deleted all queue bindings on
        the exchange are cancelled.
      </doc>

      <chassis name = "server" implement = "MUST" />
      <response name = "delete-ok" />

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "exchange" domain = "exchange-name">
        <rule name = "exists" on-failure = "not-found">
          <doc>
            The client MUST NOT attempt to delete an exchange that does not exist.
          </doc>
        </rule>
        <assert check = "notnull" />
      </field>

      <field name = "if-unused" domain = "bit" label = "delete only if unused">
        <doc>
          If set, the server will only delete the exchange if it has no queue bindings. If
          the exchange has queue bindings the server does not delete it but raises a
          channel exception instead.
        </doc>
        <rule name = "in-use" on-failure = "precondition-failed">
          <doc>
            The server MUST NOT delete an exchange that has bindings on it, if the if-unused
            field is true.
          </doc>
          <doc type = "scenario">
            The client declares an exchange, binds a queue to it, then tries to delete it
            setting if-unused to true.
          </doc>
        </rule>
      </field>

      <field name = "no-wait" domain = "no-wait" />
    </method>

    <method name = "delete-ok" synchronous = "1" index = "21"
      label = "confirm deletion of an exchange">
      <doc>This method confirms the deletion of an exchange.</doc>
      <chassis name = "client" implement = "MUST" />
    </method>
  </class>

  <!-- ==  QUEUE  ============================================================ -->

  <class name = "queue" handler = "channel" index = "50" label = "work with queues">
    <doc>
      Queues store and forward messages. Queues can be configured in the server or created at
      runtime. Queues must be attached to at least one exchange in order to receive messages
      from publishers.
    </doc>

    <doc type = "grammar">
      queue               = C:DECLARE  S:DECLARE-OK
                          / C:BIND     S:BIND-OK
                          / C:UNBIND   S:UNBIND-OK
                          / C:PURGE    S:PURGE-OK
                          / C:DELETE   S:DELETE-OK
    </doc>

    <chassis name = "server" implement = "MUST" />
    <chassis name = "client" implement = "MUST" />

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "declare" synchronous = "1" index = "10" label = "declare queue, create if needed">
      <doc>
        This method creates or checks a queue. When creating a new queue the client can
        specify various properties that control the durability of the queue and its
        contents, and the level of sharing for the queue.
      </doc>

      <rule name = "default-binding">
        <doc>
          The server MUST create a default binding for a newly-declared queue to the
          default exchange, which is an exchange of type 'direct' and use the queue
          name as the routing key.
        </doc>
        <doc type = "scenario">
          Client declares a new queue, and then without explicitly binding it to an
          exchange, attempts to send a message through the default exchange binding,
          i.e. publish a message to the empty exchange, with the queue name as routing
          key.
        </doc>
      </rule>

      <rule name = "minimum-queues">
        <doc>
          The server SHOULD support a minimum of 256 queues per virtual host and ideally,
          impose no limit except as defined by available resources.
        </doc>
        <doc type = "scenario">
          Client attempts to declare as many queues as it can until the server reports
          an error.  The resulting count must at least be 256.
        </doc>
      </rule>

      <chassis name = "server" implement = "MUST" />
      <response name = "declare-ok" />

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "queue" domain = "queue-name">
        <rule name = "default-name">
          <doc>
            The queue name MAY be empty, in which case the server MUST create a new
            queue with a unique generated name and return this to the client in the
            Declare-Ok method.
          </doc>
          <doc type = "scenario">
            Client attempts to declare several queues with an empty name. The client then
            verifies that the server-assigned names are unique and different.
          </doc>
        </rule>
        <rule name = "reserved" on-failure = "access-refused">
          <doc>
            Queue names starting with "amq." are reserved for pre-declared and
            standardised queues. The client MAY declare a queue starting with
            "amq." if the passive option is set, or the queue already exists.
          </doc>
          <doc type = "scenario">
            The client attempts to declare a non-existing queue starting with
            "amq." and with the passive option set to zero.
          </doc>
        </rule>
        <rule name = "syntax" on-failure = "precondition-failed">
          <doc>
            The queue name can be empty, or a sequence of these characters:
            letters, digits, hyphen, underscore, period, or colon.
          </doc>
          <doc type = "scenario">
            The client attempts to declare a queue with an illegal name.
          </doc>
        </rule>
      </field>

      <field name = "passive" domain = "bit" label = "do not create queue">
        <doc>
          If set, the server will reply with Declare-Ok if the queue already
          exists with the same name, and raise an error if not.  The client can
          use this to check whether a queue exists without modifying the
          server state.  When set, all other method fields except name and no-wait
          are ignored.  A declare with both passive and no-wait has no effect.
          Arguments are compared for semantic equivalence.
        </doc>
        <rule name = "passive" on-failure = "not-found">
          <doc>
            The client MAY ask the server to assert that a queue exists without
            creating the queue if not.  If the queue does not exist, the server
            treats this as a failure.
          </doc>
          <doc type = "scenario">
            Client declares an existing queue with the passive option and expects
            the server to respond with a declare-ok. Client then attempts to declare
            a non-existent queue with the passive option, and the server must close
            the channel with the correct reply-code.
          </doc>
        </rule>
        <rule name = "equivalent">
          <doc>
            If not set and the queue exists, the server MUST check that the
            existing queue has the same values for durable, exclusive, auto-delete,
            and arguments fields.  The server MUST respond with Declare-Ok if the
            requested queue matches these fields, and MUST raise a channel exception
            if not.
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
      </field>

      <field name = "durable" domain = "bit" label = "request a durable queue">
        <doc>
          If set when creating a new queue, the queue will be marked as durable. Durable
          queues remain active when a server restarts. Non-durable queues (transient
          queues) are purged if/when a server restarts. Note that durable queues do not
          necessarily hold persistent messages, although it does not make sense to send
          persistent messages to a transient queue.
        </doc>

        <rule name = "persistence">
          <doc>The server MUST recreate the durable queue after a restart.</doc>

          <doc type = "scenario">
            Client declares a durable queue. The server is then restarted. The client
            then attempts to send a message to the queue. The message should be successfully
            delivered.
          </doc>
        </rule>

        <rule name = "types">
          <doc>The server MUST support both durable and transient queues.</doc>
          <doc type = "scenario">
            A client declares two named queues, one durable and one transient.
          </doc>
        </rule>
      </field>

      <field name = "exclusive" domain = "bit" label = "request an exclusive queue">
        <doc>
          Exclusive queues may only be accessed by the current connection, and are
          deleted when that connection closes.  Passive declaration of an exclusive
          queue by other connections are not allowed.
        </doc>

        <rule name = "types">
          <doc>
            The server MUST support both exclusive (private) and non-exclusive (shared)
            queues.
          </doc>
          <doc type = "scenario">
            A client declares two named queues, one exclusive and one non-exclusive.
          </doc>
        </rule>

        <rule name = "exclusive" on-failure = "resource-locked">
          <doc>
            The client MAY NOT attempt to use a queue that was declared as exclusive
            by another still-open connection.
          </doc>
          <doc type = "scenario">
            One client declares an exclusive queue. A second client on a different
            connection attempts to declare, bind, consume, purge, delete, or declare
            a queue of the same name.
          </doc>
        </rule>
      </field>

      <field name = "auto-delete" domain = "bit" label = "auto-delete queue when unused">
        <doc>
          If set, the queue is deleted when all consumers have finished using it.  The last
          consumer can be cancelled either explicitly or because its channel is closed. If
          there was no consumer ever on the queue, it won't be deleted.  Applications can
          explicitly delete auto-delete queues using the Delete method as normal.
        </doc>

        <rule name = "pre-existence">
          <doc>
            The server MUST ignore the auto-delete field if the queue already exists.
          </doc>
          <doc type = "scenario">
            Client declares two named queues, one as auto-delete and one explicit-delete.
            Client then attempts to declare the two queues using the same names again,
            but reversing the value of the auto-delete field in each case. Verify that the
            queues still exist with the original auto-delete flag values.
          </doc>
        </rule>
      </field>

      <field name = "no-wait" domain = "no-wait" />

      <field name = "arguments" domain = "table" label = "arguments for declaration">
        <doc>
          A set of arguments for the declaration. The syntax and semantics of these
          arguments depends on the server implementation.
        </doc>
      </field>
    </method>

    <method name = "declare-ok" synchronous = "1" index = "11" label = "confirms a queue definition">
      <doc>
        This method confirms a Declare method and confirms the name of the queue, essential
        for automatically-named queues.
      </doc>

      <chassis name = "client" implement = "MUST" />

      <field name = "queue" domain = "queue-name">
        <doc>
          Reports the name of the queue. If the server generated a queue name, this field
          contains that name.
        </doc>
        <assert check = "notnull" />
      </field>

      <field name = "message-count" domain = "message-count" />

      <field name = "consumer-count" domain = "long" label = "number of consumers">
        <doc>
          Reports the number of active consumers for the queue. Note that consumers can
          suspend activity (Channel.Flow) in which case they do not appear in this count.
        </doc>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "bind" synchronous = "1" index = "20" label = "bind queue to an exchange">
      <doc>
        This method binds a queue to an exchange. Until a queue is bound it will not
        receive any messages. In a classic messaging model, store-and-forward queues
        are bound to a direct exchange and subscription queues are bound to a topic
        exchange.
      </doc>

      <rule name = "duplicates">
        <doc>
          A server MUST allow ignore duplicate bindings - that is, two or more bind
          methods for a specific queue, with identical arguments - without treating these
          as an error.
        </doc>
        <doc type = "scenario">
          A client binds a named queue to an exchange. The client then repeats the bind
          (with identical arguments).
        </doc>
      </rule>

      <rule name = "unique">
        <doc>
          A server MUST not deliver the same message more than once to a queue, even if
          the queue has multiple bindings that match the message.
        </doc>
        <doc type = "scenario">
          A client declares a named queue and binds it using multiple bindings to the
          amq.topic exchange. The client then publishes a message that matches all its
          bindings.
        </doc>
      </rule>

      <rule name = "transient-exchange">
        <doc>
          The server MUST allow a durable queue to bind to a transient exchange.
        </doc>
        <doc type = "scenario">
          A client declares a transient exchange. The client then declares a named durable
          queue and then attempts to bind the transient exchange to the durable queue.
        </doc>
      </rule>

      <rule name = "durable-exchange">
        <doc>
          Bindings of durable queues to durable exchanges are automatically durable
          and the server MUST restore such bindings after a server restart.
        </doc>
        <doc type = "scenario">
          A server declares a named durable queue and binds it to a durable exchange. The
          server is restarted. The client then attempts to use the queue/exchange combination.
        </doc>
      </rule>

      <rule name = "binding-count">
        <doc>
          The server SHOULD support at least 4 bindings per queue, and ideally, impose no
          limit except as defined by available resources.
        </doc>
        <doc type = "scenario">
          A client declares a named queue and attempts to bind it to 4 different
          exchanges.
        </doc>
      </rule>

      <chassis name = "server" implement = "MUST" />

      <response name = "bind-ok" />

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "queue" domain = "queue-name">
        <doc>Specifies the name of the queue to bind.</doc>
        <rule name = "queue-known" on-failure = "not-found">
          <doc>
            The client MUST either specify a queue name or have previously declared a
            queue on the same channel
          </doc>
          <doc type = "scenario">
            The client opens a channel and attempts to bind an unnamed queue.
          </doc>
        </rule>
        <rule name = "must-exist" on-failure = "not-found">
          <doc>
            The client MUST NOT attempt to bind a queue that does not exist.
          </doc>
          <doc type = "scenario">
            The client attempts to bind a non-existent queue.
          </doc>
        </rule>
      </field>

      <field name = "exchange" domain = "exchange-name" label = "name of the exchange to bind to">
        <rule name = "exchange-existence" on-failure = "not-found">
          <doc>
            A client MUST NOT be allowed to bind a queue to a non-existent exchange.
          </doc>
          <doc type = "scenario">
            A client attempts to bind an named queue to a undeclared exchange.
          </doc>
        </rule>
        <rule name = "default-exchange">
          <doc>
            The server MUST accept a blank exchange name to mean the default exchange.
          </doc>
          <doc type = "scenario">
            The client declares a queue and binds it to a blank exchange name.
          </doc>
        </rule>
      </field>

      <field name = "routing-key" domain = "shortstr" label = "message routing key">
        <doc>
          Specifies the routing key for the binding. The routing key is used for routing
          messages depending on the exchange configuration. Not all exchanges use a
          routing key - refer to the specific exchange documentation.  If the queue name
          is empty, the server uses the last queue declared on the channel.  If the
          routing key is also empty, the server uses this queue name for the routing
          key as well.  If the queue name is provided but the routing key is empty, the
          server does the binding with that empty routing key.  The meaning of empty
          routing keys depends on the exchange implementation.
        </doc>
        <rule name = "direct-exchange-key-matching">
          <doc>
            If a message queue binds to a direct exchange using routing key K and a
            publisher sends the exchange a message with routing key R, then the message
            MUST be passed to the message queue if K = R.
          </doc>
        </rule>
      </field>

      <field name = "no-wait" domain = "no-wait" />

      <field name = "arguments" domain = "table" label = "arguments for binding">
        <doc>
          A set of arguments for the binding. The syntax and semantics of these arguments
          depends on the exchange class.
        </doc>
      </field>
    </method>

    <method name = "bind-ok" synchronous = "1" index = "21" label = "confirm bind successful">
      <doc>This method confirms that the bind was successful.</doc>

      <chassis name = "client" implement = "MUST" />
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "unbind" synchronous = "1" index = "50" label = "unbind a queue from an exchange">
      <doc>This method unbinds a queue from an exchange.</doc>
      <rule name = "01">
        <doc>If a unbind fails, the server MUST raise a connection exception.</doc>
      </rule>
      <chassis name="server" implement="MUST"/>
      <response name="unbind-ok"/>

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "queue" domain = "queue-name">
        <doc>Specifies the name of the queue to unbind.</doc>
        <rule name = "queue-known" on-failure = "not-found">
          <doc>
            The client MUST either specify a queue name or have previously declared a
            queue on the same channel
          </doc>
          <doc type = "scenario">
            The client opens a channel and attempts to unbind an unnamed queue.
          </doc>
        </rule>
        <rule name = "must-exist" on-failure = "not-found">
          <doc>
            The client MUST NOT attempt to unbind a queue that does not exist.
          </doc>
          <doc type = "scenario">
            The client attempts to unbind a non-existent queue.
          </doc>
        </rule>
      </field>

      <field name = "exchange" domain = "exchange-name">
        <doc>The name of the exchange to unbind from.</doc>
        <rule name = "must-exist" on-failure = "not-found">
          <doc>
            The client MUST NOT attempt to unbind a queue from an exchange that
            does not exist.
          </doc>
          <doc type = "scenario">
            The client attempts to unbind a queue from a non-existent exchange.
          </doc>
        </rule>
        <rule name = "default-exchange">
          <doc>
            The server MUST accept a blank exchange name to mean the default exchange.
          </doc>
          <doc type = "scenario">
            The client declares a queue and binds it to a blank exchange name.
          </doc>
        </rule>
      </field>

      <field name = "routing-key" domain = "shortstr" label = "routing key of binding">
        <doc>Specifies the routing key of the binding to unbind.</doc>
      </field>

      <field name = "arguments" domain = "table" label = "arguments of binding">
        <doc>Specifies the arguments of the binding to unbind.</doc>
      </field>
      <!-- TODO: removed from final draft of 0.91
      <field name = "no-wait" domain = "no-wait" />
      -->
    </method>

    <method name = "unbind-ok" synchronous = "1" index = "51" label = "confirm unbind successful">
      <doc>This method confirms that the unbind was successful.</doc>
      <chassis name = "client" implement = "MUST"/>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "purge" synchronous = "1" index = "30" label = "purge a queue">
      <doc>
        This method removes all messages from a queue which are not awaiting
        acknowledgment.
      </doc>

      <rule name = "02">
        <doc>
          The server MUST NOT purge messages that have already been sent to a client
          but not yet acknowledged.
        </doc>
      </rule>

      <rule name = "03">
        <doc>
          The server MAY implement a purge queue or log that allows system administrators
          to recover accidentally-purged messages. The server SHOULD NOT keep purged
          messages in the same storage spaces as the live messages since the volumes of
          purged messages may get very large.
        </doc>
      </rule>

      <chassis name = "server" implement = "MUST" />

      <response name = "purge-ok" />

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "queue" domain = "queue-name">
        <doc>Specifies the name of the queue to purge.</doc>
        <rule name = "queue-known" on-failure = "not-found">
          <doc>
            The client MUST either specify a queue name or have previously declared a
            queue on the same channel
          </doc>
          <doc type = "scenario">
            The client opens a channel and attempts to purge an unnamed queue.
          </doc>
        </rule>
        <rule name = "must-exist" on-failure = "not-found">
          <doc>
            The client MUST NOT attempt to purge a queue that does not exist.
          </doc>
          <doc type = "scenario">
            The client attempts to purge a non-existent queue.
          </doc>
        </rule>
      </field>

      <field name = "no-wait" domain = "no-wait" />
    </method>

    <method name = "purge-ok" synchronous = "1" index = "31" label = "confirms a queue purge">
      <doc>This method confirms the purge of a queue.</doc>

      <chassis name = "client" implement = "MUST" />

      <field name = "message-count" domain = "message-count">
        <doc>
          Reports the number of messages purged.
        </doc>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "delete" synchronous = "1" index = "40" label = "delete a queue">
      <doc>
        This method deletes a queue. When a queue is deleted any pending messages are sent
        to a dead-letter queue if this is defined in the server configuration, and all
        consumers on the queue are cancelled.
      </doc>

      <rule name = "01">
        <doc>
          The server SHOULD use a dead-letter queue to hold messages that were pending on
          a deleted queue, and MAY provide facilities for a system administrator to move
          these messages back to an active queue.
        </doc>
      </rule>

      <chassis name = "server" implement = "MUST" />

      <response name = "delete-ok" />

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "queue" domain = "queue-name">
        <doc>Specifies the name of the queue to delete.</doc>
        <rule name = "queue-known" on-failure = "not-found">
          <doc>
            The client MUST either specify a queue name or have previously declared a
            queue on the same channel
          </doc>
          <doc type = "scenario">
            The client opens a channel and attempts to delete an unnamed queue.
          </doc>
        </rule>
        <rule name = "must-exist" on-failure = "not-found">
          <doc>
            The client MUST NOT attempt to delete a queue that does not exist.
          </doc>
          <doc type = "scenario">
            The client attempts to delete a non-existent queue.
          </doc>
        </rule>
      </field>

      <field name = "if-unused" domain = "bit" label = "delete only if unused">
        <doc>
          If set, the server will only delete the queue if it has no consumers. If the
          queue has consumers the server does does not delete it but raises a channel
          exception instead.
        </doc>
        <rule name = "in-use" on-failure = "precondition-failed">
          <doc>
            The server MUST NOT delete a queue that has consumers on it, if the if-unused
            field is true.
          </doc>
          <doc type = "scenario">
            The client declares a queue, and consumes from it, then tries to delete it
            setting if-unused to true.
          </doc>
        </rule>
      </field>

      <field name = "if-empty" domain = "bit" label = "delete only if empty">
        <doc>
          If set, the server will only delete the queue if it has no messages.
        </doc>
        <rule name = "not-empty" on-failure = "precondition-failed">
          <doc>
            The server MUST NOT delete a queue that has messages on it, if the
            if-empty field is true.
          </doc>
          <doc type = "scenario">
            The client declares a queue, binds it and publishes some messages into it,
            then tries to delete it setting if-empty to true.
          </doc>
        </rule>
      </field>

      <field name = "no-wait" domain = "no-wait" />
    </method>

    <method name = "delete-ok" synchronous = "1" index = "41" label = "confirm deletion of a queue">
      <doc>This method confirms the deletion of a queue.</doc>

      <chassis name = "client" implement = "MUST" />

      <field name = "message-count" domain = "message-count">
        <doc>Reports the number of messages deleted.</doc>
      </field>
    </method>
  </class>

  <!-- ==  BASIC  ============================================================ -->

  <class name = "basic" handler = "channel" index = "60" label = "work with basic content">
    <doc>
      The Basic class provides methods that support an industry-standard messaging model.
    </doc>

    <doc type = "grammar">
      basic               = C:QOS S:QOS-OK
                          / C:CONSUME S:CONSUME-OK
                          / C:CANCEL S:CANCEL-OK
                          / C:PUBLISH content
                          / S:RETURN content
                          / S:DELIVER content
                          / C:GET ( S:GET-OK content / S:GET-EMPTY )
                          / C:ACK
                          / C:REJECT
                          / C:RECOVER-ASYNC
                          / C:RECOVER S:RECOVER-OK
    </doc>

    <chassis name = "server" implement = "MUST" />
    <chassis name = "client" implement = "MAY" />

    <rule name = "01">
      <doc>
        The server SHOULD respect the persistent property of basic messages and
        SHOULD make a best-effort to hold persistent basic messages on a reliable
        storage mechanism.
      </doc>
      <doc type = "scenario">
        Send a persistent message to queue, stop server, restart server and then
        verify whether message is still present.  Assumes that queues are durable.
        Persistence without durable queues makes no sense.
      </doc>
    </rule>

    <rule name = "02">
      <doc>
        The server MUST NOT discard a persistent basic message in case of a queue
        overflow.
      </doc>
      <doc type = "scenario">
        Declare a queue overflow situation with persistent messages and verify that
        messages do not get lost (presumably the server will write them to disk).
      </doc>
    </rule>

    <rule name = "03">
      <doc>
        The server MAY use the Channel.Flow method to slow or stop a basic message
        publisher when necessary.
      </doc>
      <doc type = "scenario">
        Declare a queue overflow situation with non-persistent messages and verify
        whether the server responds with Channel.Flow or not. Repeat with persistent
        messages.
      </doc>
    </rule>

    <rule name = "04">
      <doc>
        The server MAY overflow non-persistent basic messages to persistent
        storage.
      </doc>
      <!-- Test scenario: untestable -->
    </rule>

    <rule name = "05">
      <doc>
        The server MAY discard or dead-letter non-persistent basic messages on a
        priority basis if the queue size exceeds some configured limit.
      </doc>
      <!-- Test scenario: untestable -->
    </rule>

    <rule name = "06">
      <doc>
        The server MUST implement at least 2 priority levels for basic messages,
        where priorities 0-4 and 5-9 are treated as two distinct levels.
      </doc>
      <doc type = "scenario">
        Send a number of priority 0 messages to a queue. Send one priority 9
        message.  Consume messages from the queue and verify that the first message
        received was priority 9.
      </doc>
    </rule>

    <rule name = "07">
      <doc>
        The server MAY implement up to 10 priority levels.
      </doc>
      <doc type = "scenario">
        Send a number of messages with mixed priorities to a queue, so that all
        priority values from 0 to 9 are exercised. A good scenario would be ten
        messages in low-to-high priority.  Consume from queue and verify how many
        priority levels emerge.
      </doc>
    </rule>

    <rule name = "08">
      <doc>
        The server MUST deliver messages of the same priority in order irrespective of
        their individual persistence.
      </doc>
      <doc type = "scenario">
        Send a set of messages with the same priority but different persistence
        settings to a queue.  Consume and verify that messages arrive in same order
        as originally published.
      </doc>
    </rule>

    <rule name = "09">
      <doc>
        The server MUST support un-acknowledged delivery of Basic content, i.e.
        consumers with the no-ack field set to TRUE.
      </doc>
    </rule>

    <rule name = "10">
      <doc>
        The server MUST support explicitly acknowledged delivery of Basic content,
        i.e. consumers with the no-ack field set to FALSE.
      </doc>
      <doc type = "scenario">
        Declare a queue and a consumer using explicit acknowledgements.  Publish a
        set of messages to the queue.  Consume the messages but acknowledge only
        half of them.  Disconnect and reconnect, and consume from the queue.
        Verify that the remaining messages are received.
      </doc>
    </rule>

    <!--  These are the properties for a Basic content  -->

    <!--  MIME typing -->
    <field name = "content-type"    domain = "shortstr"   label = "MIME content type" />
    <!--  MIME typing -->
    <field name = "content-encoding" domain = "shortstr"  label = "MIME content encoding" />
    <!--  For applications, and for header exchange routing -->
    <field name = "headers"         domain = "table"      label = "message header field table" />
    <!--  For queues that implement persistence -->
    <field name = "delivery-mode"   domain = "octet"      label = "non-persistent (1) or persistent (2)" />
    <!--  For queues that implement priorities -->
    <field name = "priority"        domain = "octet"      label = "message priority, 0 to 9" />
    <!--  For application use, no formal behaviour -->
    <field name = "correlation-id"  domain = "shortstr"   label = "application correlation identifier" />
    <!--  For application use, no formal behaviour but usually holds the
          name of a private response queue, when used in request messages -->
    <field name = "reply-to"        domain = "shortstr"   label = "reouting key for replies" />
    <!--  For implementation use, no formal behaviour -->
    <field name = "expiration"      domain = "shortstr"   label = "message expiration specification" />
    <!--  For application use, no formal behaviour -->
    <field name = "message-id"      domain = "shortstr"   label = "application message identifier" />
    <!--  For application use, no formal behaviour -->
    <field name = "timestamp"       domain = "timestamp"  label = "message timestamp" />
    <!--  For application use, no formal behaviour -->
    <field name = "type"            domain = "shortstr"   label = "message type name" />
    <!--  For application use, no formal behaviour -->
    <field name = "user-id"         domain = "shortstr"   label = "creating user id" />
    <!--  For application use, no formal behaviour -->
    <field name = "app-id"          domain = "shortstr"   label = "creating application id" />
    <!--  Deprecated, was old cluster-id property -->
    <field name = "reserved"        domain = "shortstr"   label = "reserved, must be empty" />

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "qos" synchronous = "1" index = "10" label = "specify quality of service">
      <doc>
        This method requests a specific quality of service. The QoS can be specified for the
        current channel or for all channels on the connection. The particular properties and
        semantics of a qos method always depend on the content class semantics. Though the
        qos method could in principle apply to both peers, it is currently meaningful only
        for the server.
      </doc>

      <chassis name = "server" implement = "MUST" />
      <response name = "qos-ok" />

      <field name = "prefetch-size" domain = "long" label = "prefetch window in octets">
        <doc>
          The client can request that messages be sent in advance so that when the client
          finishes processing a message, the following message is already held locally,
          rather than needing to be sent down the channel. Prefetching gives a performance
          improvement. This field specifies the prefetch window size in octets. The server
          will send a message in advance if it is equal to or smaller in size than the
          available prefetch size (and also falls into other prefetch limits). May be set
          to zero, meaning "no specific limit", although other prefetch limits may still
          apply. The prefetch-size is ignored if the no-ack option is set.
        </doc>
        <rule name = "01">
          <doc>
            The server MUST ignore this setting when the client is not processing any
            messages - i.e. the prefetch size does not limit the transfer of single
            messages to a client, only the sending in advance of more messages while
            the client still has one or more unacknowledged messages.
          </doc>
          <doc type = "scenario">
            Define a QoS prefetch-size limit and send a single message that exceeds
            that limit.  Verify that the message arrives correctly.
          </doc>
        </rule>
      </field>

      <field name = "prefetch-count" domain = "short" label = "prefetch window in messages">
        <doc>
          Specifies a prefetch window in terms of whole messages. This field may be used
          in combination with the prefetch-size field; a message will only be sent in
          advance if both prefetch windows (and those at the channel and connection level)
          allow it. The prefetch-count is ignored if the no-ack option is set.
        </doc>
        <rule name = "01">
          <doc>
            The server may send less data in advance than allowed by the client's
            specified prefetch windows but it MUST NOT send more.
          </doc>
          <doc type = "scenario">
            Define a QoS prefetch-size limit and a prefetch-count limit greater than
            one.  Send multiple messages that exceed the prefetch size.  Verify that
            no more than one message arrives at once.
          </doc>
        </rule>
      </field>

      <field name = "global" domain = "bit" label = "apply to entire connection">
        <doc>
          By default the QoS settings apply to the current channel only. If this field is
          set, they are applied to the entire connection.
        </doc>
      </field>
    </method>

    <method name = "qos-ok" synchronous = "1" index = "11" label = "confirm the requested qos">
      <doc>
        This method tells the client that the requested QoS levels could be handled by the
        server. The requested QoS applies to all active consumers until a new QoS is
        defined.
      </doc>
      <chassis name = "client" implement = "MUST" />
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "consume" synchronous = "1" index = "20" label = "start a queue consumer">
      <doc>
        This method asks the server to start a "consumer", which is a transient request for
        messages from a specific queue. Consumers last as long as the channel they were
        declared on, or until the client cancels them.
      </doc>

      <rule name = "01">
        <doc>
          The server SHOULD support at least 16 consumers per queue, and ideally, impose
          no limit except as defined by available resources.
        </doc>
        <doc type = "scenario">
          Declare a queue and create consumers on that queue until the server closes the
          connection. Verify that the number of consumers created was at least sixteen
          and report the total number.
        </doc>
      </rule>

      <chassis name = "server" implement = "MUST" />
      <response name = "consume-ok" />

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "queue" domain = "queue-name">
        <doc>Specifies the name of the queue to consume from.</doc>
      </field>

      <field name = "consumer-tag" domain = "consumer-tag">
        <doc>
          Specifies the identifier for the consumer. The consumer tag is local to a
          channel, so two clients can use the same consumer tags. If this field is
          empty the server will generate a unique tag.
        </doc>
        <rule name = "01" on-failure = "not-allowed">
          <doc>
            The client MUST NOT specify a tag that refers to an existing consumer.
          </doc>
          <doc type = "scenario">
            Attempt to create two consumers with the same non-empty tag, on the
            same channel.
          </doc>
        </rule>
        <rule name = "02" on-failure = "not-allowed">
          <doc>
            The consumer tag is valid only within the channel from which the
            consumer was created. I.e. a client MUST NOT create a consumer in one
            channel and then use it in another.
          </doc>
          <doc type = "scenario">
            Attempt to create a consumer in one channel, then use in another channel,
            in which consumers have also been created (to test that the server uses
            unique consumer tags).
          </doc>
        </rule>
      </field>

      <field name = "no-local" domain = "no-local" />

      <field name = "no-ack" domain = "no-ack" />

      <field name = "exclusive" domain = "bit" label = "request exclusive access">
        <doc>
          Request exclusive consumer access, meaning only this consumer can access the
          queue.
        </doc>

        <rule name = "01" on-failure = "access-refused">
          <doc>
            The client MAY NOT gain exclusive access to a queue that already has
            active consumers.
          </doc>
          <doc type = "scenario">
            Open two connections to a server, and in one connection declare a shared
            (non-exclusive) queue and then consume from the queue.  In the second
            connection attempt to consume from the same queue using the exclusive
            option.
          </doc>
        </rule>
      </field>

      <field name = "no-wait" domain = "no-wait" />

      <field name = "arguments" domain = "table" label = "arguments for declaration">
        <doc>
          A set of arguments for the consume. The syntax and semantics of these
          arguments depends on the server implementation.
        </doc>
      </field>
    </method>

    <method name = "consume-ok" synchronous = "1" index = "21" label = "confirm a new consumer">
      <doc>
        The server provides the client with a consumer tag, which is used by the client
        for methods called on the consumer at a later stage.
      </doc>
      <chassis name = "client" implement = "MUST" />
      <field name = "consumer-tag" domain = "consumer-tag">
        <doc>
          Holds the consumer tag specified by the client or provided by the server.
        </doc>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "cancel" synchronous = "1" index = "30" label = "end a queue consumer">
      <doc>
        This method cancels a consumer. This does not affect already delivered
        messages, but it does mean the server will not send any more messages for
        that consumer. The client may receive an arbitrary number of messages in
        between sending the cancel method and receiving the cancel-ok reply.
      </doc>

      <rule name = "01">
        <doc>
          If the queue does not exist the server MUST ignore the cancel method, so
          long as the consumer tag is valid for that channel.
        </doc>
        <doc type = "scenario">
          TODO.
        </doc>
      </rule>

      <chassis name = "server" implement = "MUST" />
      <response name = "cancel-ok" />

      <field name = "consumer-tag" domain = "consumer-tag" />
      <field name = "no-wait" domain = "no-wait" />
    </method>

    <method name = "cancel-ok" synchronous = "1" index = "31" label = "confirm a cancelled consumer">
      <doc>
        This method confirms that the cancellation was completed.
      </doc>
      <chassis name = "client" implement = "MUST" />
      <field name = "consumer-tag" domain = "consumer-tag" />
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "publish" content = "1" index = "40" label = "publish a message">
      <doc>
        This method publishes a message to a specific exchange. The message will be routed
        to queues as defined by the exchange configuration and distributed to any active
        consumers when the transaction, if any, is committed.
      </doc>

      <chassis name = "server" implement = "MUST" />

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "exchange" domain = "exchange-name">
        <doc>
          Specifies the name of the exchange to publish to. The exchange name can be
          empty, meaning the default exchange. If the exchange name is specified, and that
          exchange does not exist, the server will raise a channel exception.
        </doc>

        <rule name = "must-exist" on-failure = "not-found">
          <doc>
            The client MUST NOT attempt to publish a content to an exchange that
            does not exist.
          </doc>
          <doc type = "scenario">
            The client attempts to publish a content to a non-existent exchange.
          </doc>
        </rule>
        <rule name = "default-exchange">
          <doc>
            The server MUST accept a blank exchange name to mean the default exchange.
          </doc>
          <doc type = "scenario">
            The client declares a queue and binds it to a blank exchange name.
          </doc>
        </rule>
        <rule name = "02">
          <doc>
            If the exchange was declared as an internal exchange, the server MUST raise
            a channel exception with a reply code 403 (access refused).
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>

        <rule name = "03">
          <doc>
            The exchange MAY refuse basic content in which case it MUST raise a channel
            exception with reply code 540 (not implemented).
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
      </field>

      <field name = "routing-key" domain = "shortstr" label = "Message routing key">
        <doc>
          Specifies the routing key for the message. The routing key is used for routing
          messages depending on the exchange configuration.
        </doc>
      </field>

      <field name = "mandatory" domain = "bit" label = "indicate mandatory routing">
        <doc>
          This flag tells the server how to react if the message cannot be routed to a
          queue. If this flag is set, the server will return an unroutable message with a
          Return method. If this flag is zero, the server silently drops the message.
        </doc>

        <rule name = "01">
          <doc>
            The server SHOULD implement the mandatory flag.
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
      </field>

      <field name = "immediate" domain = "bit" label = "request immediate delivery">
        <doc>
          This flag tells the server how to react if the message cannot be routed to a
          queue consumer immediately. If this flag is set, the server will return an
          undeliverable message with a Return method. If this flag is zero, the server
          will queue the message, but with no guarantee that it will ever be consumed.
        </doc>

        <rule name = "01">
          <doc>
            The server SHOULD implement the immediate flag.
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
      </field>
    </method>

    <method name = "return" content = "1" index = "50" label = "return a failed message">
      <doc>
        This method returns an undeliverable message that was published with the "immediate"
        flag set, or an unroutable message published with the "mandatory" flag set. The
        reply code and text provide information about the reason that the message was
        undeliverable.
      </doc>

      <chassis name = "client" implement = "MUST" />

      <field name = "reply-code" domain = "reply-code" />
      <field name = "reply-text" domain = "reply-text" />

      <field name = "exchange" domain = "exchange-name">
        <doc>
          Specifies the name of the exchange that the message was originally published
          to.  May be empty, meaning the default exchange.
        </doc>
      </field>

      <field name = "routing-key" domain = "shortstr" label = "Message routing key">
        <doc>
          Specifies the routing key name specified when the message was published.
        </doc>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "deliver" content = "1" index = "60"
      label = "notify the client of a consumer message">
      <doc>
        This method delivers a message to the client, via a consumer. In the asynchronous
        message delivery model, the client starts a consumer using the Consume method, then
        the server responds with Deliver methods as and when messages arrive for that
        consumer.
      </doc>

      <rule name = "01">
        <doc>
          The server SHOULD track the number of times a message has been delivered to
          clients and when a message is redelivered a certain number of times - e.g. 5
          times - without being acknowledged, the server SHOULD consider the message to be
          unprocessable (possibly causing client applications to abort), and move the
          message to a dead letter queue.
        </doc>
        <doc type = "scenario">
          TODO.
        </doc>
      </rule>

      <chassis name = "client" implement = "MUST" />

      <field name = "consumer-tag" domain = "consumer-tag" />
      <field name = "delivery-tag" domain = "delivery-tag" />
      <field name = "redelivered" domain = "redelivered" />

      <field name = "exchange" domain = "exchange-name">
        <doc>
          Specifies the name of the exchange that the message was originally published to.
          May be empty, indicating the default exchange.
        </doc>
      </field>

      <field name = "routing-key" domain = "shortstr" label = "Message routing key">
        <doc>Specifies the routing key name specified when the message was published.</doc>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "get" synchronous = "1" index = "70" label = "direct access to a queue">
      <doc>
        This method provides a direct access to the messages in a queue using a synchronous
        dialogue that is designed for specific types of application where synchronous
        functionality is more important than performance.
      </doc>

      <response name = "get-ok" />
      <response name = "get-empty" />
      <chassis name = "server" implement = "MUST" />

      <!-- Deprecated: "ticket", must be zero -->
      <field name = "reserved-1" type = "short" reserved = "1" />

      <field name = "queue" domain = "queue-name">
        <doc>Specifies the name of the queue to get a message from.</doc>
      </field>
      <field name = "no-ack" domain = "no-ack" />
    </method>

    <method name = "get-ok" synchronous = "1" content = "1" index = "71"
      label = "provide client with a message">
      <doc>
        This method delivers a message to the client following a get method. A message
        delivered by 'get-ok' must be acknowledged unless the no-ack option was set in the
        get method.
      </doc>

      <chassis name = "client" implement = "MAY" />

      <field name = "delivery-tag" domain = "delivery-tag" />
      <field name = "redelivered" domain = "redelivered" />
      <field name = "exchange" domain = "exchange-name">
        <doc>
          Specifies the name of the exchange that the message was originally published to.
          If empty, the message was published to the default exchange.
        </doc>
      </field>

      <field name = "routing-key" domain = "shortstr" label = "Message routing key">
        <doc>Specifies the routing key name specified when the message was published.</doc>
      </field>

      <field name = "message-count" domain = "message-count" />
    </method>

    <method name = "get-empty" synchronous = "1" index = "72"
      label = "indicate no messages available">
      <doc>
        This method tells the client that the queue has no messages available for the
        client.
      </doc>
      <chassis name = "client" implement = "MAY" />
      <!-- Deprecated: "cluster-id", must be empty -->
      <field name = "reserved-1" type = "shortstr" reserved = "1" />
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "ack" index = "80" label = "acknowledge one or more messages">
      <doc>
        This method acknowledges one or more messages delivered via the Deliver or Get-Ok
        methods. The client can ask to confirm a single message or a set of messages up to
        and including a specific message.
      </doc>

      <chassis name = "server" implement = "MUST" />

      <field name = "delivery-tag" domain = "delivery-tag" />
      <field name = "multiple" domain = "bit" label = "acknowledge multiple messages">
        <doc>
          If set to 1, the delivery tag is treated as "up to and including", so that the
          client can acknowledge multiple messages with a single method. If set to zero,
          the delivery tag refers to a single message. If the multiple field is 1, and the
          delivery tag is zero, tells the server to acknowledge all outstanding messages.
        </doc>
        <rule name = "exists" on-failure = "precondition-failed">
          <doc>
            The server MUST validate that a non-zero delivery-tag refers to a delivered
            message, and raise a channel exception if this is not the case.  On a transacted
            channel, this check MUST be done immediately and not delayed until a Tx.Commit.
            Specifically, a client MUST not acknowledge the same message more than once.
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "reject" index = "90" label = "reject an incoming message">
      <doc>
        This method allows a client to reject a message. It can be used to interrupt and
        cancel large incoming messages, or return untreatable messages to their original
        queue.
      </doc>

      <rule name = "01">
        <doc>
          The server SHOULD be capable of accepting and process the Reject method while
          sending message content with a Deliver or Get-Ok method. I.e. the server should
          read and process incoming methods while sending output frames. To cancel a
          partially-send content, the server sends a content body frame of size 1 (i.e.
          with no data except the frame-end octet).
        </doc>
      </rule>

      <rule name = "02">
        <doc>
          The server SHOULD interpret this method as meaning that the client is unable to
          process the message at this time.
        </doc>
        <doc type = "scenario">
          TODO.
        </doc>
      </rule>

      <rule name = "03">
        <doc>
          The client MUST NOT use this method as a means of selecting messages to process.
        </doc>
        <doc type = "scenario">
          TODO.
        </doc>
      </rule>

      <chassis name = "server" implement = "MUST" />

      <field name = "delivery-tag" domain = "delivery-tag" />

      <field name = "requeue" domain = "bit" label = "requeue the message">
        <doc>
          If requeue is true, the server will attempt to requeue the message.  If requeue
          is false or the requeue  attempt fails the messages are discarded or dead-lettered.
        </doc>

        <rule name = "01">
          <doc>
            The server MUST NOT deliver the message to the same client within the
            context of the current channel. The recommended strategy is to attempt to
            deliver the message to an alternative consumer, and if that is not possible,
            to move the message to a dead-letter queue. The server MAY use more
            sophisticated tracking to hold the message on the queue and redeliver it to
            the same client at a later stage.
          </doc>
          <doc type = "scenario">
            TODO.
          </doc>
        </rule>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "recover-async" index = "100" label = "redeliver unacknowledged messages"
        deprecated = "1">
      <doc>
        This method asks the server to redeliver all unacknowledged messages on a
        specified channel. Zero or more messages may be redelivered.  This method
        is deprecated in favour of the synchronous Recover/Recover-Ok.
      </doc>
      <rule name = "01">
        <doc>
          The server MUST set the redelivered flag on all messages that are resent.
        </doc>
        <doc type = "scenario">
          TODO.
        </doc>
      </rule>
      <chassis name = "server" implement = "MAY" />
      <field name = "requeue" domain = "bit" label = "requeue the message">
        <doc>
          If this field is zero, the message will be redelivered to the original
          recipient. If this bit is 1, the server will attempt to requeue the message,
          potentially then delivering it to an alternative subscriber.
        </doc>
      </field>
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "recover" index = "110" label = "redeliver unacknowledged messages">
      <doc>
        This method asks the server to redeliver all unacknowledged messages on a
        specified channel. Zero or more messages may be redelivered.  This method
        replaces the asynchronous Recover.
      </doc>
      <rule name = "01">
        <doc>
          The server MUST set the redelivered flag on all messages that are resent.
        </doc>
        <doc type = "scenario">
          TODO.
        </doc>
      </rule>
      <chassis name = "server" implement = "MUST" />
      <field name = "requeue" domain = "bit" label = "requeue the message">
        <doc>
          If this field is zero, the message will be redelivered to the original
          recipient. If this bit is 1, the server will attempt to requeue the message,
          potentially then delivering it to an alternative subscriber.
        </doc>
      </field>
    </method>

    <method name = "recover-ok" synchronous = "1" index = "111" label = "confirm recovery">
      <doc>
        This method acknowledges a Basic.Recover method.
      </doc>
      <chassis name = "client" implement = "MUST" />
    </method>
  </class>

  <!-- ==  TX  =============================================================== -->

  <class name = "tx" handler = "channel" index = "90" label = "work with transactions">
    <doc>
      The Tx class allows publish and ack operations to be batched into atomic
      units of work.  The intention is that all publish and ack requests issued
      within a transaction will complete successfully or none of them will.
      Servers SHOULD implement atomic transactions at least where all publish
      or ack requests affect a single queue.  Transactions that cover multiple
      queues may be non-atomic, given that queues can be created and destroyed
      asynchronously, and such events do not form part of any transaction.
      Further, the behaviour of transactions with respect to the immediate and
      mandatory flags on Basic.Publish methods is not defined.
    </doc>

    <rule name = "not multiple queues">
      <doc>
      Applications MUST NOT rely on the atomicity of transactions that
      affect more than one queue.
      </doc>
    </rule>
    <rule name = "not immediate">
      <doc>
      Applications MUST NOT rely on the behaviour of transactions that
      include messages published with the immediate option.
      </doc>
    </rule>
    <rule name = "not mandatory">
      <doc>
      Applications MUST NOT rely on the behaviour of transactions that
      include messages published with the mandatory option.
      </doc>
    </rule>

    <doc type = "grammar">
      tx                  = C:SELECT S:SELECT-OK
                          / C:COMMIT S:COMMIT-OK
                          / C:ROLLBACK S:ROLLBACK-OK
    </doc>

    <chassis name = "server" implement = "SHOULD" />
    <chassis name = "client" implement = "MAY" />

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "select" synchronous = "1" index = "10" label = "select standard transaction mode">
      <doc>
        This method sets the channel to use standard transactions. The client must use this
        method at least once on a channel before using the Commit or Rollback methods.
      </doc>
      <chassis name = "server" implement = "MUST" />
      <response name = "select-ok" />
    </method>

    <method name = "select-ok" synchronous = "1" index = "11" label = "confirm transaction mode">
      <doc>
        This method confirms to the client that the channel was successfully set to use
        standard transactions.
      </doc>
      <chassis name = "client" implement = "MUST" />
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "commit" synchronous = "1" index = "20" label = "commit the current transaction">
      <doc>
        This method commits all message publications and acknowledgments performed in
        the current transaction.  A new transaction starts immediately after a commit.
      </doc>
      <chassis name = "server" implement = "MUST" />
      <response name = "commit-ok" />

      <rule name = "transacted" on-failure = "precondition-failed">
        <doc>
          The client MUST NOT use the Commit method on non-transacted channels.
        </doc>
        <doc type = "scenario">
          The client opens a channel and then uses Tx.Commit.
        </doc>
      </rule>
    </method>

    <method name = "commit-ok" synchronous = "1" index = "21" label = "confirm a successful commit">
      <doc>
        This method confirms to the client that the commit succeeded. Note that if a commit
        fails, the server raises a channel exception.
      </doc>
      <chassis name = "client" implement = "MUST" />
    </method>

    <!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->

    <method name = "rollback" synchronous = "1" index = "30"
      label = "abandon the current transaction">
      <doc>
        This method abandons all message publications and acknowledgments performed in
        the current transaction. A new transaction starts immediately after a rollback.
        Note that unacked messages will not be automatically redelivered by rollback;
        if that is required an explicit recover call should be issued.
      </doc>
      <chassis name = "server" implement = "MUST" />
      <response name = "rollback-ok" />

      <rule name = "transacted" on-failure = "precondition-failed">
        <doc>
          The client MUST NOT use the Rollback method on non-transacted channels.
        </doc>
        <doc type = "scenario">
          The client opens a channel and then uses Tx.Rollback.
        </doc>
      </rule>
    </method>

    <method name = "rollback-ok" synchronous = "1" index = "31" label = "confirm successful rollback">
      <doc>
        This method confirms to the client that the rollback succeeded. Note that if an
        rollback fails, the server raises a channel exception.
      </doc>
      <chassis name = "client" implement = "MUST" />
    </method>
  </class>

</amqp>
Add a New Comment

Edit | Files | Tags | Source | Print

rating: 0+x

Recent chatter

Zyre, a new RestMS implementation
On 2008/12/25, iMatix released Zyre, a new implementation of RestMS. Zyre is a web service that proxies HTTP to AMQP, implementing the RestMS specification. Zyre includes its own fully configurable web server. Download from openamq.org.

AMQP-0-9-1 XML
This is the final editorial draft 5 of AMQP0-91.xml. Please comment on this page, or by email to all participants of the 0-9-1 SIG.

New IP policy
We're writing a new IP policy designed to be simpler and compatible with IETF and the AMQP Working Group. Get involved, if you have an opinion!

Background to DMP
From www.openamq.org, a backgrounder to DMP and the OpenAMQ implementation of it ("Direct Mode"). Explains why iMatix designed DMP, and why it makes message transfer 5-6x faster.

AMQP presentation
An introduction to AMQP made by Alexis Richardson from Rabbit Technologies.