aboutsummaryrefslogtreecommitdiffstats
path: root/lib/diameter/doc
diff options
context:
space:
mode:
Diffstat (limited to 'lib/diameter/doc')
-rw-r--r--lib/diameter/doc/src/notes.xml70
1 files changed, 70 insertions, 0 deletions
diff --git a/lib/diameter/doc/src/notes.xml b/lib/diameter/doc/src/notes.xml
index 50f568abaa..60478606ad 100644
--- a/lib/diameter/doc/src/notes.xml
+++ b/lib/diameter/doc/src/notes.xml
@@ -43,6 +43,76 @@ first.</p>
<!-- ===================================================================== -->
+<section><title>diameter 2.0</title>
+
+ <section><title>Improvements and New Features</title>
+ <list>
+ <item>
+ <p>
+ Let candidate peers be passed to diameter:call/4</p>
+ <p>
+ With call option peer, to allow a request to be sent to a
+ peer that hasn't advertised support for the application
+ in question.</p>
+ <p>
+ RFC 6733 2.4 requires a node to send the application
+ identifiers of all locally supported applications at
+ capabilities exchange, but not all nodes respect this for
+ the common application, and diameter itself will send
+ D[WP][RA] without the common application having been
+ explicitly advertised. Regarding the common application
+ as implicit renders Result-Code 5010
+ (DIAMETER_NO_COMMON_APPLICATION) meaningless however, so
+ allow any request to be sent as long as there is a
+ configured dictionary to support it.</p>
+ <p>
+ Own Id: OTP-14338</p>
+ </item>
+ <item>
+ <p>
+ Improve performance of message encode/decode and related
+ handling.</p>
+ <p>
+ Dictionaries using @custom_types or @codecs will need to
+ adapt the corresponding functions to accept an additional
+ argument that is now passed through encode/decode, which
+ was required to remove various process dictionary-based
+ workarounds that have been used to solve problems in the
+ past.</p>
+ <p>
+ *** POTENTIAL INCOMPATIBILITY ***</p>
+ <p>
+ Own Id: OTP-14343</p>
+ </item>
+ <item>
+ <p>
+ Add transport options to avoid deadlock and allow for
+ load regulation.</p>
+ <p>
+ Both diameter_tcp and diameter_sctp now accept two new
+ configuration options: sender and message_cb. The former
+ causes outgoing sends to take place in a dedicated
+ process, to avoid the possibility of deadlock when both
+ the transport process and its peer block in send. The
+ latter allows a callback to control the reading of
+ messages on the socket, to allow for backpressure towards
+ peers when the rate of incoming traffic is greater than
+ can otherwise be handled.</p>
+ <p>
+ Neither of these options are yet documented, but are
+ unlikely to change unless problems are discovered. The
+ sender option is not the default since it should probably
+ always be used in combination with message_cb, to prevent
+ incoming requests from being read at a higher rate than a
+ peer allows outgoing answers to be sent.</p>
+ <p>
+ Own Id: OTP-14455 Aux Id: ERL-332 </p>
+ </item>
+ </list>
+ </section>
+
+</section>
+
<section><title>diameter 1.12.2</title>
<section><title>Fixed Bugs and Malfunctions</title>