[Shtoom] recent changes ; planning for 0.3
Anthony Baxter
anthony at interlink.com.au
Sat Sep 25 08:34:07 CEST 2004
So I've recently completed a fairly major set of reworkings
of the RTP/SDP handling and the audio device handling. In
addition, I've put in a simple minded playout buffer that
should hopefully stop the creeping delay problem that many
people have complained about - it will add a 30ms delay to
incoming calls. In addition, speex support is checked in,
although it's not been tested against any other clients
yet, but I believe I've got it right. I will be checking in support
for a bunch of the G.72x codecs in the next week, after I
test them against cisco's implementation.
So, what's left before 0.3 release? Here's the entries from
TODO:
Definite before 0.3:
- make sure Doug is up-to-date with recent refactorings
- all interfaces must be current and correct
- Asterisk is rejecting our ACK - e.g. to 613 at fwd.pulver.com
- doug: implement leg bridging (inbound and outbound)
- doug: use dropCall()'s deferred to generate correct events
- doug: s.d.conferencing's entrance soundburst bug
- ui(qt): hide multiple call support for now
- Some sort of "debug dump" command that dumps relevant info (ui,
audio, &c) for bugreports.
- package for Mac, Windows
- phone: when on a call, decline a second incoming call?
- documentation, documentation, documentation
- document Doug API
- update website
- create shtoom wiki
- set svn:executable on the scripts
- set up a decent roundup instance, load all entries from TODO
Maybe before 0.3:
- G.72x codec
- doug: placeCall to a broken address doesn't give generate a
correct event
- let doug handle wav and au files natively.
- MediaLayer: a more sophisticated playout buffer
- cocoa native UI into it's own thread to workaround broken
cfreactor
- stun: timeout the STUN call, generate a nice error
- sip: retry registration failures (e.g. for testcall)
- sip: move more stuff into the Dialog class
- ui(tk): new dialogs that don't block, and have a timeout.
- sip/call: the state should become an object, encapsulating the
current state, and the state machine. This makes it _much_
easier to handle retransmits and the like.
- RTCP + NAT == RFC 3605
If there's other things you'd like to see, now's the time to speak up.
Anthony
--
Anthony Baxter <anthony at interlink.com.au>
It's never too late to have a happy childhood.
_______________________________________________
Divmod-dev mailing list
Divmod-dev at divmod.org
http://divmod.org/users/mailman.twistd/listinfo/divmod-dev
More information about the Shtoom
mailing list