summaryrefslogtreecommitdiff
path: root/src/call-channel.c
AgeCommit message (Collapse)AuthorFilesLines
2010-09-21various Call classes: use tp_clear_object, tp_clear_pointerSimon McVittie1-3/+1
2010-08-24Don't leak Call channels when waiting for caps.Will Thompson1-2/+4
call_channel_capabilities_discovered_cb() gets the channel by calling g_simple_async_result_get_source_object(), which returns a reference to the source object. Previously, it then failed to release its reference. This meant that if we have to wait for caps before completing a channel request, we'd always leak the channel. There's no regression test here. I found it while porting GabbleBaseCallChannel to use TpBaseChannel, because the latter holds a ref to its connection, and so leaking the Call channel leaks the connection which prevents it falling off the bus, which makes the existing regression tests for waiting for caps both fail! I've also audited Wocky and fixed a few places that make this same mistake there.
2010-08-11Replace all JingleSessionState with JingleStateSimon McVittie1-3/+3
As a side benefit, this fixes some cases of comparing a JingleState to a JingleSessionState (using the fact that they happen to be numerically equal), which gcc 4.5 apparently warns about.
2010-04-29Call channel is assume to keep a ref to the internal sessionSjoerd Simons1-1/+1
2010-04-23Make the user adding content in a call a virtual functionSjoerd Simons1-0/+32
2010-04-05Add some debugging if we don't understand the media typesSjoerd Simons1-1/+2
2010-04-01Merge remote branch 'upstream/master' into muc-callsSjoerd Simons1-0/+4
Conflicts: src/call-channel.c
2010-03-26Fix refcounting of the JingleSessionSjoerd Simons1-0/+6
When asking the jingle factory to create a session it gives you back a object you don't have a ref to, so first of all make sure we have a ref. And second, do actually unref it when disposing the channel :)
2010-02-18Wait for capabilities before completely initializing a call.Mike Ruprecht1-14/+142
If Gabble doesn't have a contact's capabilities, it should wait until they are retrieved before returning from the CreateChannel or EnsureChannel call. The caps must be checked to ensure they have the required capabilities before returning whether or not the call was successful. Some of the capabilities checking code was copied from media-channel.c.
2010-02-18Remove CallContent when CallMemberContent is removed.Mike Ruprecht1-0/+24
This is for CallChannels only. CallMucChannels shouldn't have their CallContents removed when there are no members left. Someone might join the conference.
2010-02-18CallChannel: Handle additional contents added by the remote side.Mike Ruprecht1-0/+35
2010-02-18Use a virtual function for the actual acceptance of callsSjoerd Simons1-0/+10
2010-02-18Purely use CallMemberContents instead of JingleContentsSjoerd Simons1-8/+14
2010-02-18Start using seperate CallMember objectsSjoerd Simons1-32/+127
2010-02-18Make the handle type a class variableSjoerd Simons1-0/+4
2010-02-18Split a BaseCallChannel out of CallChannelSjoerd Simons1-945/+20
Both Muji related call channels and 1-to-1 call channel should be able to share quite some code. The main difference is in setting up the call. This is a very minimal split, more 1-to-1 call specific code will move out of the base class into the call-channel
2010-01-22Merge remote branch 'maiku/replace-codecoffers'Sjoerd Simons1-1/+1
Conflicts: src/call-content.c src/call-content.h
2010-01-22Merge remote branch 'maiku/handle-new-content'Sjoerd Simons1-0/+28
2010-01-20Add a _deinit function to CallContent.Mike Ruprecht1-1/+1
Using _deinit, CallContent will wait for its CodecOffer (if any) to be freed before allowing itself to be freed.
2010-01-20Add debug messages for unexpected events in Accept().Mike Ruprecht1-0/+3
2010-01-20In Accept(), set sending state from Pending to Sending automatically.Mike Ruprecht1-0/+2
As per the spec, in Accept(), all streams that are in the pending send state and are in contents whose disposition is initial should be automatically set to the sending state.
2010-01-20Handle when the remote end adds a content to the session.Mike Ruprecht1-0/+26
2010-01-20Set the Ringing_Locally flag when Ringing is called.Mike Ruprecht1-2/+29
2010-01-20Ensure Accept isn't called in an invalid state.Mike Ruprecht1-0/+16
2010-01-20Return an error if AddContent is called on an ended Call.Mike Ruprecht1-0/+8
2010-01-06Represent bare jids as resource = NULL, not resource = ""Simon McVittie1-5/+3
The RFCs don't explicitly say anywhere that the empty string is not a valid resource... also, it's cleaner this way. However, this does mean changing a couple of function signatures. Also move the definition of SESSION_MAP_KEY_FORMAT close to its only use.
2010-01-05GabbleCallChannel, GabbleMediaChannel: improve debug for existing sessionsSimon McVittie1-2/+9
2009-12-24Free all data in finalizeSjoerd Simons1-10/+3
2009-12-24Implement of CallMembersSjoerd Simons1-0/+56
2009-12-24Correctly implement mutable contentsSjoerd Simons1-2/+2
2009-12-24Add a name argument to the jingle sessionSjoerd Simons1-1/+1
Allow other bits of code to suggest a name when creating a new content in a jingle session. Also default audio/video streams to have respectively Audio/Video as their default names. In case of content naming colisions suffix the suggested/default name with _%d to prevent confusing clients that think content names are unique.
2009-12-24Implement Name, Disposition and Creator properties on Call Content objectsSjoerd Simons1-4/+15
2009-12-24Implement the AddContent methodSjoerd Simons1-0/+42
2009-12-24Use the first transport of an incoming stream as the default transportSjoerd Simons1-0/+5
2009-12-24Make call_channel_create_content return the object path or an errorSjoerd Simons1-7/+19
2009-12-24Simple implementation of HangupSjoerd Simons1-0/+24
2009-12-24Implement CallStateSjoerd Simons1-0/+64
2009-12-24Get all call state properties directly from private fieldsSjoerd Simons1-19/+27
2009-12-24Implement dummy versions of various propertiesSjoerd Simons1-0/+63
2009-12-24Implement the HardwareStreaming propertySjoerd Simons1-0/+12
2009-12-24Indentation triviaSjoerd Simons1-4/+4
2009-12-24Implement dummy version of RingingSjoerd Simons1-0/+9
2009-12-24Fix various coding style and types, also remove some dead codeSjoerd Simons1-1/+1
2009-12-23Fix identationSjoerd Simons1-1/+1
2009-12-23Use G_PARAM_STATIC_STRINGS where applicableSjoerd Simons1-2/+1
2009-12-23Fix indentationSjoerd Simons1-33/+33
2009-12-08Update the Call spec to the 0.19.0 versionSjoerd Simons1-2/+2
2009-12-07Impement CallChannel.AcceptSjoerd Simons1-0/+20
2009-12-07Refactor CallContent creation so it works correctly for incoming callsSjoerd Simons1-25/+62
2009-12-07Use constructed instead of constructorSjoerd Simons1-10/+5