[svn-commits] oej: branch group/astridevcon2007 r65831 - /team/group/astridevcon2007/

svn-commits at lists.digium.com svn-commits at lists.digium.com
Thu May 24 07:06:26 MST 2007


Author: oej
Date: Thu May 24 09:06:25 2007
New Revision: 65831

URL: http://svn.digium.com/view/asterisk?view=rev&rev=65831
Log:
Adding steven's notes

Added:
    team/group/astridevcon2007/iaxmeeting.txt   (with props)
Modified:
    team/group/astridevcon2007/general-agenda.txt

Modified: team/group/astridevcon2007/general-agenda.txt
URL: http://svn.digium.com/view/asterisk/team/group/astridevcon2007/general-agenda.txt?view=diff&rev=65831&r1=65830&r2=65831
==============================================================================
--- team/group/astridevcon2007/general-agenda.txt (original)
+++ team/group/astridevcon2007/general-agenda.txt Thu May 24 09:06:25 2007
@@ -111,6 +111,9 @@
 
 - Your topic here
 
+- Commit time - present issues for the bug tracker and let a committer 
+
+
 Short talks
 ===========
 

Added: team/group/astridevcon2007/iaxmeeting.txt
URL: http://svn.digium.com/view/asterisk/team/group/astridevcon2007/iaxmeeting.txt?view=auto&rev=65831
==============================================================================
--- team/group/astridevcon2007/iaxmeeting.txt (added)
+++ team/group/astridevcon2007/iaxmeeting.txt Thu May 24 09:06:25 2007
@@ -1,0 +1,89 @@
+RFC Review
+
+Covered Previously
+
+UTF8 Support
+
+An Asterisk-wide issue.  Primarily due to the inability of the
+Asterisk string handling
+code does not handle embedded nulls.  This must be addressed
+throughout the Asterisk.
+
+Video/Code Negotiation
+
+IAX2 and the Asterisk codec management system has run out of unique
+bit patterns to
+represent codecs.  The system supports a total of 16 possible values
+(duh).  There are
+a number of methods of handling this but any method selected will
+require complete
+backwards compatibility.
+
+SteveK suggested setting ALL bits to indicate that codec negotiation
+is being deferred
+to a codec negotiation process based on data set in IEs.
+
+Another issue: no way to handle keyframes for video.  The standard
+doesn't explicitly
+describe video packetization.
+
+Renegotiation: we currently have a defacto means of changing the codec
+by simply sending
+a full frame with the new codec bit set.  We need to work out a formal
+method for handling
+renegotiation.  The call should maintain the capabilities of the
+remote party and allow
+either party to switch the format to any of the supported formats.
+
+What are the semantics for the codec negotiation?  The order of the
+FORMAT IEs represents
+the order of preference?  How do the endpoints produce a
+"counterproposal" for the other
+parties proposal?
+
+In mid call, how do we renegotiate?  We will likely need a new frame
+type to implement
+this.  It will steal many of the semantics of the NEW frame's codec
+negotiation but
+won't actually set up a new call. RENEW?  FORMATS? Can we use the key
+rotation mechanism
+from the encryption code to provide this?  Oops.  Doesn't look like we
+have one.  We
+can use the same new frametype to do both.  Can we use ACCEPT to
+respond to the RENEW?
+Can we use INVAL to reject the proposal if it won't work?
+
+How can we handle multiple streams?  Example: 2 audio streams or 2
+video streams.  (Think
+the Brady Bunch method of video conferencing).  How about all streams
+in a conference
+going to all endpoint (which would then handle the mixing).  There are
+issues with
+packetization related to handling video.  Sequence numbers are
+required (rather than just
+using timestamps) because some video may break up a single codec-level
+block of data
+across a series of packets.
+
+Russell: The current video frame definition is broken.  It needs to be
+altered/expanded
+to provide for the full range of video options.
+
+Implementation
+
+Much of the issues with Impl. have been fixed with the new
+multi-threaded impl.  Some
+of the lookup of calls could be better handled by hashes rather than
+the current array
+method.  This causes the thread to lock and unlock each call as it
+searches.  Russell
+has some code to improve this, but it is not ready for prime time.
+
+Still need to cover:
+
+- Presence & Events
+- Encryption
+- IPv6
+
+-------------------------------------
+--Steven Sokol

Propchange: team/group/astridevcon2007/iaxmeeting.txt
------------------------------------------------------------------------------
    svn:eol-style = native

Propchange: team/group/astridevcon2007/iaxmeeting.txt
------------------------------------------------------------------------------
    svn:keywords = Author Date Id Revision

Propchange: team/group/astridevcon2007/iaxmeeting.txt
------------------------------------------------------------------------------
    svn:mime-type = text/plain



More information about the svn-commits mailing list