path: root/lib/diameter/doc/src/notes.xml

<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd" [
  <!ENTITY % also SYSTEM "seealso.ent" >
  <!ENTITY % here SYSTEM "seehere.ent" >
  %also;
  %here;
]>

<chapter>

<header>
<copyright>
<year>2011</year>
<year>2014</year>
<holder>Ericsson AB. All Rights Reserved.</holder>
</copyright>
<legalnotice>
The contents of this file are subject to the Erlang Public License,
Version 1.1, (the "License"); you may not use this file except in
compliance with the License. You should have received a copy of the
Erlang Public License along with this software. If not, it can be
retrieved online at http://www.erlang.org/.

Software distributed under the License is distributed on an "AS IS"
basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
the License for the specific language governing rights and limitations
under the License.

</legalnotice>

<title>Release Notes</title>
<prepared></prepared>
<docno></docno>
<date></date>
<rev></rev>
<file>notes.xml</file>
</header>

<p>
Releases are listed in reverse chronological order, most recent
first.</p>

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

<section><title>diameter 1.7</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Improve robustness.</p>
          <p>
	    Counters returned by diameter:service_info/2 now only
	    count messages known to the dictionary in question, so
	    that an attacker cannot cause arbitrarily many counters
	    to be created.</p>
          <p>
	    Messages to the Erlang log have been minimized, and those
	    related to traffic have been removed entirely since an
	    attacker could cause a node to be logged to death.
	    Consequently, the default answer_errors configuration has
	    been changed from report to discard. A service needs to
	    be restarted for the change in default to take effect.</p>
          <p>
	    Own Id: OTP-11721</p>
        </item>
        <item>
          <p>
	    Fix request table leak.</p>
          <p>
	    Outgoing Diameter requests are stored in a table until an
	    answer is received or times out. Calling
	    diameter:stop_service/1 before this took place would
	    orphan the entries, resulting in a memory leak.</p>
          <p>
	    Own Id: OTP-11893</p>
        </item>
        <item>
          <p>
	    Fix broken SCTP transport.</p>
          <p>
	    OTP-11593 caused the sending of answer messages over SCTP
	    to fail.</p>
          <p>
	    Own Id: OTP-11901 Aux Id: OTP-11593 </p>
        </item>
        <item>
          <p>
	    Fix watchdog process leak.</p>
          <p>
	    A failed capabilities exchange on a listening transport
	    would orphan a process, causing a memory leak.</p>
          <p>
	    Own Id: OTP-11934</p>
        </item>
        <item>
          <p>
	    Fix incorrect handling of incoming DPR.</p>
          <p>
	    In the case of a listening transport, a reconnection by a
	    peer following DPR could transition the watchdog state to
	    REOPEN instead of OKAY.</p>
          <p>
	    Own Id: OTP-11938</p>
        </item>
        <item>
          <p>
	    Fix handling of AVP length errors on unknown AVPs.</p>
          <p>
	    An AVP (Header) length that pointed past the end of the
	    message was not flagged as a 5014 error in this case.
	    Moreover, encoding such an AVP in the Failed-AVP of an
	    answer message as a consequence of other errors (eg.
	    M-bit, resulting in 5001) failed if the AVP contained a
	    complete header.</p>
          <p>
	    Dictionary files must be recompiled for the fix to have
	    effect.</p>
          <p>
	    Own Id: OTP-11946</p>
        </item>
        <item>
          <p>
	    Fix broken check in dictionary compilation.</p>
          <p>
	    That an AVP specified in the content of a @codecs or
	    @custom_types section was undefined went undetected,
	    causing compilation to fail when attempting to lookup the
	    AVP's type.</p>
          <p>
	    Own Id: OTP-11958</p>
        </item>
      </list>
    </section>


    <section><title>Improvements and New Features</title>
      <list>
        <item>
          <p>
	    Add result code counters for CEA, DWA, and DPA.</p>
          <p>
	    In addition to the existing result code counters on other
	    answer messages.</p>
          <p>
	    Own Id: OTP-11891</p>
        </item>
        <item>
          <p>
	    Add best-effort decode of AVPs within Failed-AVP.</p>
          <p>
	    OTP-11007 disabled the decode of AVPs in Failed-AVP since
	    errors could cause the decode of Failed-AVP itself to
	    fail. Component AVPs are now decoded if possible,
	    otherwise not. AVPs of type Grouped are decoded as much
	    as possible, as deeply as possible.</p>
          <p>
	    Dictionary files must be recompiled for the fix to have
	    effect.</p>
          <p>
	    Own Id: OTP-11936 Aux Id: OTP-11007 </p>
        </item>
        <item>
          <p>
	    Add counters for encode errors in outgoing Diameter
	    messages.</p>
          <p>
	    In addition to the existing counters on decode errors.
	    The latter now count independently of result codes in
	    answer messages since decode errors do not preclude the
	    presence of a result code.</p>
          <p>
	    Own Id: OTP-11937</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.6</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Add missing check at dictionary compilation.</p>
          <p>
	    In particular, that an AVP defined as having type Grouped
	    in an @avp_types section has a corresponding definition
	    in a @grouped section.</p>
          <p>
	    Own Id: OTP-11561</p>
        </item>
        <item>
          <p>
	    Correct documentation on the setting of Origin-State-Id</p>
          <p>
	    It was incorrectly stated that the AVP would be set in an
	    outgoing DPR/DPA.</p>
          <p>
	    Own Id: OTP-11583</p>
        </item>
        <item>
          <p>
	    Change interface for communicating outbound stream id to
	    diameter_sctp</p>
          <p>
	    The module uses the transport_data field of record
	    diameter_packet to communicate the stream on which the an
	    incoming message is received and on which an outgoing
	    message should be sent, the previous interface being that
	    both are communicated as a tuple of the form {stream,
	    Id}. However, since diameter retains the value of an
	    incoming request's transport_data unless the
	    corresponding answer message specifies otherwise, the
	    behaviour in this case is to send an answer on the
	    outbound stream with the same identifier as the that of
	    the inbound stream on which the request was received. If
	    the inbound stream id is greater than or equal to the
	    number of outbound streams then this is guaranteed to
	    fail, causing the transport process in question to
	    terminate. There is no relationship between inbound and
	    outbound stream identifiers so diameter_sctp's imposition
	    of one is simply wrong.</p>
          <p>
	    Outbound stream ids are now communicated with a different
	    tuple: {outstream, Id}, interpreted modulo the number of
	    outbound streams. Thus, retention of an inbound request's
	    transport_data has no effect on the selection of an
	    outbound stream.</p>
          <p>
	    The change in interface is not strictly backwards
	    compatible because of the new atom for the outbound
	    stream. However, as there is currently no documented way
	    of obtaining the available number of outbound streams for
	    a peer connection, there is no way for a client to have
	    known the range of ids from which it could reliably have
	    chosen with the previous interface, so any setting of the
	    outbound stream has probably been unintentional. Not
	    explicitly specifying an outbound stream now results in a
	    round-robin selection.</p>
          <p>
	    Thanks to Sharmila Pillai for reporting the problem.</p>
          <p>
	    *** POTENTIAL INCOMPATIBILITY ***</p>
          <p>
	    Own Id: OTP-11593</p>
        </item>
        <item>
          <p>
	    Fix unicode path failure in diameter_make:codec/2.</p>
          <p>
	    A dictionary path containing a unicode codepoint > 255
	    caused the function to fail.</p>
          <p>
	    Own Id: OTP-11655</p>
        </item>
        <item>
          <p>
	    Fix 'accept' config to diameter_sctp.</p>
          <p>
	    OTP-10893 added support for {accept, Match} tuples to
	    specify addresses or regexps that should be matched
	    against peer addresses to decide whether or not a newly
	    established association should be retained, but this
	    hasn't been functional in the SCTP case because of
	    missing support in inet(3).</p>
          <p>
	    The display of both local and peer addresses in
	    diameter:service_info/2 output has also been corrected.</p>
          <p>
	    Own Id: OTP-11661 Aux Id: OTP-10229 </p>
        </item>
        <item>
          <p>
	    Be lenient with the M-bit in Grouped AVPs.</p>
          <p>
	    RFC 6733 says this, in 4.4:</p>
          <p>
	    <taglist><item><p><c>Receivers of a Grouped AVP that does
	    not have the 'M' (mandatory) bit set and one or more of
	    the encapsulated AVPs within the group has the 'M'
	    (mandatory) bit set MAY simply be ignored if the Grouped
	    AVP itself is unrecognized. The rule applies even if the
	    encapsulated AVP with its 'M' (mandatory) bit set is
	    further encapsulated within other sub-groups, i.e., other
	    Grouped AVPs embedded within the Grouped
	    AVP.</c></p></item></taglist></p>
          <p>
	    The first sentence is mangled but take it to mean this:</p>
          <p>
	    <taglist><item><p><c>An unrecognized AVP of type Grouped
	    that does not set the 'M' bit MAY be ignored even if one
	    of its encapsulated AVPs sets the 'M'
	    bit.</c></p></item></taglist></p>
          <p>
	    This is a bit of a non-statement since if the AVP is
	    unrecognized then its type is unknown. We therefore don't
	    know that its data bytes contain encapsulated AVPs, so
	    can't but ignore any of those that set the M-bit. Doing
	    anything else when the type *is* known would be
	    inconsistent.</p>
          <p>
	    OTP-11087 (R16B03) caused the M-bit on any unrecognized
	    AVP to be regarded as an error, unrecognized being taken
	    to mean "not explicitly defined as a member of its
	    container". (That is, an AVP that can't be packed into a
	    dedicated record field, which is slightly stronger than
	    "not defined".) This fixed the original intention for
	    top-level AVPs but broke the required leniency for
	    Grouped AVPs whose type is known. This leniency is now
	    restored.</p>
          <p>
	    Note that dictionary files need to be recompiled for the
	    change to have effect.</p>
          <p>
	    Thanks to Rory McKeown for reporting the problem.</p>
          <p>
	    Own Id: OTP-11675 Aux Id: OTP-11087 </p>
        </item>
        <item>
          <p>
	    Fix pick_peer case clause failure.</p>
          <p>
	    In the case of {call_mutates_state, true} configuration
	    on the service in question, any peer selection that
	    failed to select a peer resulted in a case clause
	    failure. This was noticed in the case of a peer failover
	    in which an alternate peer wasn't available.</p>
          <p>
	    Own Id: OTP-11789</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.5</title>

    <section><title>Improvements and New Features</title>
      <list>
        <item>
          <p>
	    Rename reconnect_timer to connect_timer.</p>
          <p>
	    The former is still accepted for backwards compatibility,
	    but the name is misleading given the semantics of the
	    timer.</p>
          <p>
	    Own Id: OTP-11168</p>
        </item>
        <item>
          <p>
	    Extend diameter_make(3).</p>
          <p>
	    Dictionaries can now be compiled from strings, not just
	    filesystem paths, and results can be returned instead of
	    written to the filesystem.</p>
          <p>
	    Own Id: OTP-11348</p>
        </item>
        <item>
          <p>
	    Remove hardcoding of diameter_base as @prefix on
	    dictionaries for application id 0.</p>
          <p>
	    Own Id: OTP-11361</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.4.4</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Fix setting of End-to-End and Hop-by-Hop Identifiers in
	    outgoing DWA.</p>
          <p>
	    Broken by OTP-11184, which caused the identifiers to be
	    set anew, discarding the values from the incoming DWR.</p>
          <p>
	    Own Id: OTP-11367</p>
        </item>
        <item>
          <p>
	    Fix handling of 5014, DIAMETER_INVALID_AVP_LENGTH.</p>
          <p>
	    The error was detected as 5004,
	    DIAMETER_INVALID_AVP_VALUE, for some Diameter types, in
	    which case an AVP length that pointed past the end of a
	    message resulted in encode failure.</p>
          <p>
	    Own Id: OTP-11395</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.4.3</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Fix UTF8String encode.</p>
          <p>
	    Encode now accepts any nested list of codepoints and
	    binaries. A list containing a binary was previously
	    misinterpreted and the documentation was incomplete.</p>
          <p>
	    Own Id: OTP-11172</p>
        </item>
        <item>
          <p>
	    Ensure DWR isn't sent immediately after DWA.</p>
          <p>
	    This was possible if the timing was unfortunate. An
	    incoming DWR now properly resets the watchdog timer.</p>
          <p>
	    Own Id: OTP-11184</p>
        </item>
        <item>
          <p>
	    Fix faulty encode of Failed-AVP</p>
          <p>
	    Reception of a CER, DWR or DPR that has decode failures
	    caused encode of the corresponding answer message to
	    fail.</p>
          <p>
	    Own Id: OTP-11293</p>
        </item>
        <item>
          <p>
	    Fix broken service_opt() spawn_opt.</p>
          <p>
	    The option was ignored.</p>
          <p>
	    Own Id: OTP-11299</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.4.2</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Fix handling of 5014 (INVALID_AVP_LENGTH) errors.</p>
          <p>
	    This was in some cases reported as 3009
	    (INVALID_AVP_BITS).</p>
          <p>
	    Note that the correction is partially implemented in
	    modules generated by diameterc(1): a dictionary file must
	    be recompiled for the correction to apply to any messages
	    it defines.</p>
          <p>
	    Own Id: OTP-11007</p>
        </item>
        <item>
          <p>
	    Fix faulty capitalization in release notes.</p>
          <p>
	    Diameter = the protocol.<br/> diameter = the Erlang
	    application.</p>
          <p>
	    Own Id: OTP-11014</p>
        </item>
        <item>
          <p>
	    Fix watchdog memory leak.</p>
          <p>
	    Entries were not removed from a service-specific ets
	    table, causing them to be orphaned at connection
	    reestablishment for listening transports, and
	    diameter:remove_transport/2 for both listening and
	    connecting transports.</p>
          <p>
	    The fault was introduced by OTP-10692 in diameter-1.4.1
	    (R16B).</p>
          <p>
	    Own Id: OTP-11019 Aux Id: OTP-10692 </p>
        </item>
        <item>
          <p>
	    Fix decode failure on AVP Length &lt; 8.</p>
          <p>
	    The failure caused the message in question to be
	    discarded.</p>
          <p>
	    Own Id: OTP-11026</p>
        </item>
        <item>
          <p>
	    Respect Host-IP-Address configuration.</p>
          <p>
	    Addresses returned from a transport module were always
	    used to populate Host-IP-Address AVP's in an outgoing
	    CER/CEA, which precluded the sending of a VIP address.
	    Transport addresses are now only used if Host-IP-Address
	    is unspecified.</p>
          <p>
	    Own Id: OTP-11045</p>
        </item>
        <item>
          <p>
	    Fix mkdir race.</p>
          <p>
	    Install could fail if examples/code and examples/dict
	    were created in parallel. Noticed on FreeBSD.</p>
          <p>
	    Own Id: OTP-11051</p>
        </item>
        <item>
          <p>
	    Fix recognition of 5001 on mandatory AVP's.</p>
          <p>
	    An AVP setting the M-bit was not regarded as erroneous if
	    it was defined in the dictionary in question and its
	    container (message or Grouped AVP) had an 'AVP' field.
	    It's now regarded as a 5001 error (AVP_UNSUPPORTED), as
	    in the case that the AVP is not defined.</p>
          <p>
	    Note that the correction is partially implemented in
	    modules generated by diameterc(1): a dictionary file must
	    be recompiled for the correction to apply to any messages
	    it defines.</p>
          <p>
	    Own Id: OTP-11087</p>
        </item>
        <item>
          <p>
	    Fix setting of Failed-AVP on handle_request
	    {answer_message, 5xxx} return.</p>
          <p>
	    Failed-AVP was never in the outgoing answer-message. It
	    is now set with the AVP from the first entry with the
	    specified Result-Code in the errors field of the incoming
	    diameter_packet, if found.</p>
          <p>
	    Own Id: OTP-11092</p>
        </item>
        <item>
          <p>
	    Fix watchdog function_clause</p>
          <p>
	    A listening transport on a service that allowed multiple
	    connections to the same peer could result in a
	    function_clause error in module diameter_watchdog. The
	    resulting crash was harmless but unseemly.</p>
          <p>
	    Thanks to Aleksander Nycz.</p>
          <p>
	    Own Id: OTP-11115</p>
        </item>
        <item>
          <p>
	    Fix population of Failed-AVP.</p>
          <p>
	    In cases in which diameter populated this AVP, many
	    values were sent instead of one as suggested by RFC 6733.
	    This was partially corrected by OTP-11007.</p>
          <p>
	    Own Id: OTP-11127 Aux Id: OTP-11007 </p>
        </item>
        <item>
          <p>
	    Fix list-valued Vendor-Specific-Application-Id config</p>
          <p>
	    R16B (specifically, OTP-10760) broke the handling of such
	    configuration, resulting in a function clause error if
	    the list was not of length 3, and faulty interpretation
	    of the list's contents otherwise. Only record-valued
	    configuration was properly interpreted.</p>
          <p>
	    Own Id: OTP-11165</p>
        </item>
      </list>
    </section>


    <section><title>Improvements and New Features</title>
      <list>
        <item>
          <p>
	    Allow peer connections to be shared between Erlang nodes
	    for the purpose of sending outgoing requests.</p>
          <p>
	    A diameter_app(3) pick_peer/4 callback gets a list of
	    remote candidates as argument, allowing a callback on one
	    node to select a transport connection established on
	    another node. The service_opt() share_peers controls the
	    extent to which local connections are shared with remote
	    nodes. The service_opt() use_shared_peers controls the
	    extent to which connections shared from remote nodes are
	    utilized on the local node.</p>
          <p>
	    Own Id: OTP-9610</p>
        </item>
        <item>
          <p>
	    Allow listening diameter_{tcp,sctp} transports to be
	    configured with remote addresses.</p>
          <p>
	    Option 'accept' allows remote addresses to be configured
	    as tuples or regular expressions. Remote addresses are
	    matched against the configured values at connection
	    establishment, any non-matching address causing the
	    connection to be aborted.</p>
          <p>
	    Own Id: OTP-10893</p>
        </item>
        <item>
          <p>
	    Detect more transport_opt() configuration errors at
	    diameter:add_transport/2.</p>
          <p>
	    Many errors would previously not be detected until
	    transport start, diameter:add_transport/2 returning 'ok'
	    but transport connections failing to be established. An
	    error tuple is now returned.</p>
          <p>
	    Own Id: OTP-10972</p>
        </item>
        <item>
          <p>
	    Make explicit local address configuration optional in
	    diameter_tcp:start/3.</p>
          <p>
	    The default address (as determined by gen_tcp) is now
	    used when a local address is not explicitly configured.</p>
          <p>
	    Own Id: OTP-10986</p>
        </item>
        <item>
          <p>
	    Improve handling of unrecognized service options.</p>
          <p>
	    Such options were silently ignored by
	    diameter:start_service/2. An error tuple is now returned.</p>
          <p>
	    Own Id: OTP-11017</p>
        </item>
        <item>
          <p>
	    Don't send default Inband-Security-Id in CER/CEA.</p>
          <p>
	    RFC 6733 recommends against the use of
	    Inband-Security-Id. Only send a value that differs from
	    the default, NO_INBAND_SECURITY = 0.</p>
          <p>
	    Own Id: OTP-11050</p>
        </item>
        <item>
          <p>
	    Make spawn options for request processes configurable.</p>
          <p>
	    Own Id: OTP-11060</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.4.1.1</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Fix broken Vendor-Specific-Application-Id configuration.</p>
          <p>
	    RFC 6733 changed the definition of this Grouped AVP,
	    changing the arity of Vendor-Id from 1* to 1. A component
	    Vendor-Id can now be either list- or integer-valued in
	    service and transport configuration, allowing it to be
	    used with both RFC 3588 and RFC 6733 dictionaries.</p>
          <p>
	    Own Id: OTP-10942</p>
        </item>
      </list>
    </section>


    <section><title>Improvements and New Features</title>
      <list>
        <item>
          <p>
	    Add transport_opt() watchdog_config to allow non-standard
	    behaviour of the watchdog state machine.</p>
          <p>
	    This can be useful during test but should not be used on
	    nodes that must conform to RFC 3539.</p>
          <p>
	    Own Id: OTP-10898</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.4.1</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Fix erroneous watchdog transition from DOWN to INITIAL.</p>
          <p>
	    This transition took place when a peer connection was
	    reestablished following a failed capabilities exchange.
	    RFC 3539 requires DOWN to transition into REOPEN.</p>
          <p>
	    Own Id: OTP-10692</p>
        </item>
      </list>
    </section>


    <section><title>Improvements and New Features</title>
      <list>
        <item>
          <p>
	    Add application_opt() request_errors to make the handling
	    of incoming requests containing decode errors
	    configurable.</p>
          <p>
	    The value 'callback' ensures that a handle_request
	    callback takes place for all such requests, the default
	    being for diameter to answer 3xxx series errors itself.</p>
          <p>
	    Own Id: OTP-10686</p>
        </item>
        <item>
          <p>
	    Add transport_opt() length_errors.</p>
          <p>
	    The value determines how messages received over the
	    transport interface with an incorrect Message Length are
	    dealt with.</p>
          <p>
	    Own Id: OTP-10687</p>
        </item>
        <item>
          <p>
	    Add commentary on RFC 6733 to Standards Compliance
	    chapter of the User's Guide.</p>
          <p>
	    Own Id: OTP-10688</p>
        </item>
        <item>
          <p>
	    Allow a 5xxx result code in an answer-message on peer
	    connections using the RFC 6733 common dictionary.</p>
          <p>
	    RFC 6733 allows this while RFC 3588 does not. A
	    handle_request callback can return {answer_message,
	    3000..3999|5000..5999} in the simplest case.</p>
          <p>
	    Own Id: OTP-10759</p>
        </item>
        <item>
          <p>
	    Add dictionaries for RFC 6733.</p>
          <p>
	    Both the common and accounting dictionaries differ from
	    their RFC 3588 counterparts, which is reflected in
	    generated record definitions. Application configuration
	    on a service or transport determines the dictionary that
	    will be used on a given peer connection.</p>
          <p>
	    Own Id: OTP-10760</p>
        </item>
        <item>
          <p>
	    Allow a handle_request callback to control diameter's
	    setting of Result-Code and Failed-AVP.</p>
          <p>
	    Setting errors = false in a returned #diameter_packet{}
	    disables the setting.</p>
          <p>
	    Own Id: OTP-10761</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.4</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Add registered server names to the app file.</p>
          <p>
	    Own Id: OTP-10442</p>
        </item>
        <item>
          <p>
	    Fix #diameter_header{} handling broken by OTP-10445.</p>
          <p>
	    The fault caused the the header of a [Header | Avps]
	    request to be ignored if both end_to_end_id and
	    hop_by_hop_id were undefined.</p>
          <p>
	    Own Id: OTP-10609</p>
        </item>
        <item>
          <p>
	    Fix error handling for handle_request callback.</p>
          <p>
	    A callback that returned a #diameter_packet{} would fail
	    if the incoming request had decode errors.</p>
          <p>
	    Own Id: OTP-10614</p>
        </item>
        <item>
          <p>
	    Fix timing of service start event.</p>
          <p>
	    The event did not necessarily precede other events as
	    documented.</p>
          <p>
	    Own Id: OTP-10618</p>
        </item>
        <item>
          <p>
	    Fix setting of header T flag at peer failover.</p>
          <p>
	    The flag is now set in the diameter_header record passed
	    to a prepare_retransmit callback.</p>
          <p>
	    Own Id: OTP-10619</p>
        </item>
        <item>
          <p>
	    Fix sending of CER/CEA timeout event at capx_timeout.</p>
          <p>
	    The event was not sent as documented.</p>
          <p>
	    Own Id: OTP-10628</p>
        </item>
        <item>
          <p>
	    Fix improper setting of Application-ID in the Diameter
	    header of an answer message whose E flag is set.</p>
          <p>
	    The value should be that of the request in question. The
	    fault caused it always to be 0.</p>
          <p>
	    Own Id: OTP-10655</p>
        </item>
        <item>
          <p>
	    Fix faulty handling of AVP length errors.</p>
          <p>
	    An incorrect AVP length but no other errors caused an
	    incoming request to fail.</p>
          <p>
	    Own Id: OTP-10693</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.3.1</title>

    <section><title>Known Bugs and Problems</title>
      <list>
        <item>
          <p>
	    Fix function clause resulting from use of an eval
	    callback.</p>
          <p>
	    Own Id: OTP-10685</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.3</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Fix faulty handling of Origin-State-Id and faulty config
	    values.</p>
          <p>
	    The former was expected in a list despite the
	    documentation requiring (correctly) an integer. A bare
	    value for a list-valued capability was not handled.</p>
          <p>
	    Own Id: OTP-10440</p>
        </item>
        <item>
          <p>
	    Fix timing of up/down events.</p>
          <p>
	    Previously, a call to diameter:call/4 following a peer_up
	    callback might incorrectly return {error, no_connection},
	    depending on timing. Both events now follow the
	    corresponding callbacks.</p>
          <p>
	    Own Id: OTP-10459</p>
        </item>
        <item>
          <p>
	    Make diameter:service_info/2 usable in peer_up, peer_down
	    and pick_peer callbacks.</p>
          <p>
	    Except for in pick_peer when {call_mutates_state, false},
	    it would previously hang indefinitely.</p>
          <p>
	    Own Id: OTP-10460</p>
        </item>
        <item>
          <p>
	    Verify that End-to-End and Hop-by-Hop Identifiers in an
	    incoming CEA/DPA match those sent in the corresponding
	    CER/DPR.</p>
          <p>
	    The values were previously ignored. Answers whose
	    identifiers do not match are handled as unexpected.</p>
          <p>
	    Own Id: OTP-10565</p>
        </item>
        <item>
          <p>
	    Fix formatting problems in PDF documentation.</p>
          <p>
	    In particular, text corresponding to links in HTML was
	    omitted in preformatted blocks. There are still issues
	    with indentation but this is not diameter-specific.</p>
          <p>
	    Own Id: OTP-10583</p>
        </item>
      </list>
    </section>


    <section><title>Improvements and New Features</title>
      <list>
        <item>
          <p>
	    Let prepare_request, prepare_retransmit and
	    handle_request callbacks return a function to be invoked
	    on outgoing messages after encode.</p>
          <p>
	    This allows encoded messages to be logged for example.</p>
          <p>
	    Own Id: OTP-10441</p>
        </item>
        <item>
          <p>
	    Add service_opt() 'restrict_connections' to allow
	    multiple transport connections with the same peer.</p>
          <p>
	    Own Id: OTP-10443</p>
        </item>
        <item>
          <p>
	    Add service_opt() 'sequence' to allow the masking of a
	    constant onto the topmost bits of End-to-End and
	    Hop-by-Hop identifiers.</p>
          <p>
	    This allows the same service on different nodes to use
	    distinct values in outgoing request messages.</p>
          <p>
	    Own Id: OTP-10445</p>
        </item>
        <item>
          <p>
	    Add diameter:service_info(PeerRef) to return the
	    transport_ref() and transport_opt() list of the
	    corresponding transport.</p>
          <p>
	    This allows easy access to these from diameter_app
	    callbacks that only get peer_ref() as an argument.</p>
          <p>
	    Own Id: OTP-10470</p>
        </item>
        <item>
          <p>
	    Add reference pages diameter_codec(3) and
	    diameter_make(3).</p>
          <p>
	    Own Id: OTP-10471</p>
        </item>
        <item>
          <p>
	    Add events for service start and stop.</p>
          <p>
	    Own Id: OTP-10492</p>
        </item>
        <item>
          <p>
	    Add transport_opt() 'disconnect_cb' to make the sending
	    of DPR configurable.</p>
          <p>
	    Whether or not DPR should be sent at application stop,
	    service stop or transport removal is determined by the
	    value returned by the callback, as is the
	    Disconnect-Cause and timeout if DPA is not received.</p>
          <p>
	    Own Id: OTP-10493</p>
        </item>
        <item>
          <p>
	    Add transport_opt() 'capx_timeout' for the timeout
	    associated with non-reception of CER/CEA.</p>
          <p>
	    Own Id: OTP-10554</p>
        </item>
        <item>
          <p>
	    Allow a handle_request callback to return a
	    #diameter_packet{}.</p>
          <p>
	    This allows an answer to set transport_data and header
	    fields.</p>
          <p>
	    Own Id: OTP-10566</p>
        </item>
        <item>
          <p>
	    Update documentation for RFC 6733.</p>
          <p>
	    RFC 3588 is now obsolete.</p>
          <p>
	    Own Id: OTP-10568</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.2</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Fix broken Result-Code setting and Destination-Host/Realm
	    extraction.</p>
          <p>
	    Result-Code was assumed to have arity 1 when setting this
	    value in an answer to a request containing AVP decode
	    errors. Destination-Host/Realm were only correctly
	    extracted from messages in the common application.</p>
          <p>
	    Own Id: OTP-10202</p>
        </item>
        <item>
          <p>
	    Handle insufficient capabilities configuration more
	    gracefully.</p>
          <p>
	    A transport that does not have sufficient capabilities
	    configuration in order to encode CER/CEA will now emit an
	    error report noting the configuration error and exit
	    instead of failing. The error is not detected at
	    diameter:add_transport/2 since there is no requirement
	    that a service be configured before its transports.</p>
          <p>
	    Own Id: OTP-10203</p>
        </item>
        <item>
          <p>
	    Ensure a failing peer_up/down callback does not affect
	    transport connections to other peers.</p>
          <p>
	    Such a failure would previously have taken down all of a
	    service's connections.</p>
          <p>
	    Own Id: OTP-10215</p>
        </item>
      </list>
    </section>


    <section><title>Improvements and New Features</title>
      <list>
        <item>
          <p>
	    Statistics related to Diameter messages can be retrieved
	    using diameter:service_info/2.</p>
          <p>
	    Both Diameter and socket-level statistics are available,
	    for both incoming and outgoing messages.</p>
          <p>
	    Own Id: OTP-9608</p>
        </item>
        <item>
          <p>
	    Allow multiple transport_module/config to
	    diameter:add_transport/2.</p>
          <p>
	    Multiple values are attempted in sequence until one
	    results in an established connection. This provides a way
	    for a connecting transport to specify configuration in
	    order of preference. (For example, SCTP before TCP.)</p>
          <p>
	    Own Id: OTP-9885</p>
        </item>
        <item>
          <p>
	    Add events for state transitions in the RFC 3539 watchdog
	    state machine.</p>
          <p>
	    The watchdog state is also available through
	    diameter:service_info/2.</p>
          <p>
	    Own Id: OTP-10212</p>
        </item>
        <item>
          <p>
	    Add diameter:service_info(SvcName, connections).</p>
          <p>
	    This provides an alternative to
	    diameter:service_info(SvcName, transport) that presents
	    information per established connection instead of per
	    transport reference.</p>
          <p>
	    Own Id: OTP-10213</p>
        </item>
        <item>
          <p>
	    Assorted documentation corrections/improvements.</p>
          <p>
	    Own Id: OTP-10216</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.1</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Fix fault in sending of 'closed' events.</p>
          <p>
	    The fault made it possible for the 'closed' event not to
	    be sent following a failed capabilities exchange.</p>
          <p>
	    Own Id: OTP-9824</p>
        </item>
        <item>
          <p>
	    Fix faulty diameterc -name/-prefix.</p>
          <p>
	    A minor blunder when introducing the new dictionary
	    parser in diameter-1.0 broke these options.</p>
          <p>
	    Own Id: OTP-9826</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 1.0</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Fix faulty cleanup after diameter:remove_transport/2.</p>
          <p>
	    Removing a transport removed the configuration but did
	    not prevent the transport process from being restarted.</p>
          <p>
	    Own Id: OTP-9756</p>
        </item>
      </list>
    </section>


    <section><title>Improvements and New Features</title>
      <list>
        <item>
          <p>
	    Add support for TLS over TCP.</p>
          <p>
	    RFC 3588 requires that a Diameter server support TLS. In
	    practice this seems to mean TLS over SCTP since there are
	    limitations with running over SCTP: see RFC 6083 (DTLS
	    over SCTP), which is a response to RFC 3436 (TLS over
	    SCTP). The current RFC 3588 draft acknowledges this by
	    equating TLS with TLS/TCP and DTLS/SCTP.</p>
          <p>
	    TLS handshaking can take place either following a CER/CEA
	    that negotiates TLS using the Inband-Security-Id AVP (the
	    method documented in RFC 3588) or immediately following
	    connection establishment (the method added to the current
	    draft).</p>
          <p>
	    Own Id: OTP-9605</p>
        </item>
        <item>
          <p>
	    Improvements to the dictionary parser.</p>
          <p>
	    The dictionary parser now emits useful error messages in
	    case of faults in the input file, also identifying the
	    line number at which the fault was detected. There are
	    semantic checks that were missing in the previous parser,
	    a fault in the interpretation of vendor id's in
	    combination with @inherits has been fixed and @end can be
	    used to terminate parsing explicitly instead of always
	    parsing to end of file.</p>
          <p>
	    Own Id: OTP-9639</p>
        </item>
        <item>
          <p>
	    Improve dictionary reusability.</p>
          <p>
	    Reusing a dictionary just to get a different generated
	    module name or prefix previously required taking a copy
	    of the source, which may consist of several files if
	    inheritance is used, just to edit a couple of lines which
	    don't affect the semantics of the Diameter application
	    being defined. Options --name, --prefix and --inherits
	    have been added to diameterc to allow corresponding
	    values to be set at compile time.</p>
          <p>
	    Own Id: OTP-9641</p>
        </item>
        <item>
          <p>
	    Add capabilities_cb transport option.</p>
          <p>
	    Its value is a function that's applied to the transport
	    reference and capabilities record after capabilities
	    exchange. If a callback returns anything but 'ok' then
	    the connection is closed. In the case of an incoming CER,
	    the callback can return a result code with which to
	    answer. Multiple callbacks can be specified and are
	    applied until either all return 'ok' or one doesn't.</p>
          <p>
	    This provides a way to reject a peer connection.</p>
          <p>
	    Own Id: OTP-9654</p>
        </item>
        <item>
          <p>
	    Add @codecs to dictionary format.</p>
          <p>
	    The semantics are similar to @custom_types but results in
	    codec functions of the form TypeName(encode|decode,
	    AvpName, Data) rather than AvpName(encode|decode,
	    TypeName, Data). That is, the role of the AVP name and
	    Diameter type name are reversed. This eliminates the need
	    for exporting one function for each AVP sharing a common
	    specialized encode/decode.</p>
          <p>
	    Own Id: OTP-9708 Aux Id: OTP-9639 </p>
        </item>
        <item>
          <p>
	    Add #diameter_callback{} for more flexible callback
	    configuration.</p>
          <p>
	    The record allows individual functions to be configured
	    for each of the diameter_app(3) callbacks, as well as a
	    default callback.</p>
          <p>
	    Own Id: OTP-9777</p>
        </item>
      </list>
    </section>

</section>

<section><title>diameter 0.10</title>

    <section><title>Fixed Bugs and Malfunctions</title>
      <list>
        <item>
          <p>
	    Handle #sctp_paddr_change and #sctp_pdapi_event from
	    gen_sctp.</p>
          <p>
	    The events are enabled by default but diameter_sctp
	    neither disabled nor dealt with them. Reception of such
	    an event caused a transport process to crash.</p>
          <p>
	    Own Id: OTP-9538</p>
        </item>
        <item>
          <p>
	    Fix header folding bug.</p>
          <p>
	    A prepare_request callback from diameter can return a
	    diameter_header record in order to set values in the
	    header of an outgoing request. A fault in
	    diameter_lib:fold_tuple/3 caused the subsequent encode of
	    the outgoing request to fail.</p>
          <p>
	    Own Id: OTP-9577</p>
        </item>
        <item>
          <p>
	    Fix bugs in sending of answer-message replies.</p>
          <p>
	    3001 (DIAMETER_COMMAND_UNSUPPORTED) was not sent since
	    the decode placed the AVP list in the wrong field of the
	    diameter_packet, causing the subsequent encode to fail.
	    Session-Id was also set improperly, causing encode to
	    fail even in this case.</p>
          <p>
	    Own Id: OTP-9578</p>
        </item>
        <item>
          <p>
	    Fix improper use of error_logger:info_report/2.</p>
          <p>
	    Function doesn't take a format string and arguments as it
	    was called. Instead use error_logger:info_report/1 and
	    use the same report format as used for warning and error
	    reports.</p>
          <p>
	    Own Id: OTP-9579</p>
        </item>
        <item>
          <p>
	    Fix and clarify semantics of peer filters.</p>
          <p>
	    An eval filter returning a non-true value caused the call
	    process to fail and the doc was vague on how an exception
	    was treated. Clarify that the non-tuple host/realm
	    filters assume messages of a certain form.</p>
          <p>
	    Own Id: OTP-9580</p>
        </item>
        <item>
          <p>
	    Fix and clarify relay behaviour.</p>
          <p>
	    Implicit filtering of the sending peer in relaying a
	    request could cause loop detection to be preempted in a
	    manner not specified by RFC3588. Reply with 3002
	    (DIAMETER_UNABLE_TO_DELIVER) on anything but an answer to
	    a relayed request.</p>
          <p>
	    Own Id: OTP-9583</p>
        </item>
      </list>
    </section>


    <section><title>Improvements and New Features</title>
      <list>
        <item>
          <p>
	    @id required in dictionary files only when @messages is
	    specified.</p>
          <p>
	    @id defines an application identifier and this is used
	    only when sending or receiving messages. A dictionary can
	    define only AVP's however, to be included by other
	    dictionaries using @inherits, in which case it makes no
	    sense to require @id.</p>
          <p>
	    Note that message definitions are not inherited with
	    @inherits, only AVP's</p>
          <p>
	    Own Id: OTP-9467</p>
        </item>
        <item>
          <p>
	    Allow @enum when AVP is defined in an inherited
	    dictionary.</p>
          <p>
	    3GPP standards (for one) extend the values allowed for
	    RFC 3588 AVP's of type Enumerated. Previously, extending
	    an AVP was only possible by completely redefining the
	    AVP.</p>
          <p>
	    Own Id: OTP-9469</p>
        </item>
        <item>
          <p>
	    Migrate testsuites to pure common test and add both
	    suites and testcases.</p>
          <p>
	    Own Id: OTP-9553</p>
        </item>
        <item>
          <p>
	    Requests of arbitrary form.</p>
          <p>
	    diameter:call/4 can be passed anything, as long as the
	    subsequent prepare_request callback returns a term that
	    can be encoded.</p>
          <p>
	    Own Id: OTP-9581</p>
        </item>
      </list>
    </section>

</section>

<section>
<title>diameter 0.9</title>

<p>
Initial release of the diameter application.</p>

<p>
Known issues or limitations:</p>

<list>

<item>
<p>
Some agent-related functionality is not entirely complete.
In particular, support for proxy agents, that advertise specific
Diameter applications but otherwise relay messages in much the same
way as relay agents (for which a handle_request
callback can return a <c>relay</c> tuple), will be completed in an
upcoming release.
There may also be more explicit support for redirect agents, although
redirect behaviour can be implemented with the current
functionality.</p>

</item>

<item>
<p>
There is some asymmetry in the treatment of messages sent as
<c>diameter_header/avp</c> records and those sent in the "normal"
fashion, and not all of this is documented.
This is related to the previous point since this form of sending a
message was introduced specifically to handle relay agent behaviour
using the same callback interface as for client/server behaviour.</p>
</item>

<item>
<p>
The User's Guide is currently quite thin.
The introductory chapter followed by the examples (in the application
<c>examples</c> subdirectory) may be sufficient
for those having some familiarity with the Diameter protocol but the
intention is to provide more introductory text.
The reference documentation is quite complete, although some points
could likely be expanded upon.</p>
</item>

<item>
<p>
The function diameter:service_info/2
can be used to retrieve information about a started service
(statistics, information about connected peers, etc) but
this is not yet documented and both the input and output may change
in the next release.</p>
</item>


</list>

<p>
See <seealso marker="diameter_soc">Standards Compliance</seealso> for
standards-related issues.</p>
</section>

</chapter>