- 01 Oct, 2010 2 commits
-
-
-
Youness Alaoui authored
-
- 29 Sep, 2010 1 commit
-
-
Youness Alaoui authored
The agent->local_addresses should not change when we gather_candidates, let the user set it with nice_agent_add_local_address, if he does not, then we should auto-discover interface *during* the gathering process, and not save it. Currently, when we call gather_candidates, we discover interfaces and save it to agent->local_addresses as if someone called the _add_local_address API... problem is that if we have 2/3 interfaces, we do an audio call.. an interface disappears (vmware closed/whatever) then we upgrade to a video call, then libnice will reuse the old interface list instead of the current ones..
-
- 23 Aug, 2010 3 commits
-
-
Youness Alaoui authored
-
Youness Alaoui authored
-
Youness Alaoui authored
-
- 05 Aug, 2010 1 commit
-
-
Youness Alaoui authored
-
- 20 Jul, 2010 2 commits
-
-
Olivier Crête authored
-
Olivier Crête authored
-
- 16 Jul, 2010 2 commits
-
-
Youness Alaoui authored
wtf.. should not set added to -1 since we continue adding other candidates and e want to schedule_next if one succeeded
-
Youness Alaoui authored
-
- 21 May, 2010 8 commits
-
-
Olivier Crête authored
-
Olivier Crête authored
-
Olivier Crête authored
-
Olivier Crête authored
-
Olivier Crête authored
-
Olivier Crête authored
-
Olivier Crête authored
-
Olivier Crête authored
-
- 19 May, 2010 1 commit
-
-
Youness Alaoui authored
-
- 08 May, 2010 1 commit
-
-
Sjoerd Simons authored
We need to keep the GMainContext referenced so we can be sure it's still alive when we remove the sources we've attached to it.
-
- 16 Feb, 2010 17 commits
-
-
Youness Alaoui authored
After 5 hours of debugging... finally! remember to always remove the weak pointers when not needed anymore!
-
Youness Alaoui authored
-
Youness Alaoui authored
-
Youness Alaoui authored
Use weak pointers to make sure that the recv callback does not destroy our object, and if it does, make sure not to use the agent or PseudoTcpSocket to avoid a crash
-
Youness Alaoui authored
Fix a attach/detach recv bug where if we detach our recv callback from the nice agent during the callback of the tcp_readable, we would end up having the loop consume all data from PseudoTcpSocket without notifying anyone of it. This now only consumes data if the callback is available, and calls the readable callback when we attach a new recv callback if the pseudo tcp socket was still readable and all its data wasn't consumed yet
-
Youness Alaoui authored
-
Youness Alaoui authored
-
Youness Alaoui authored
-
Youness Alaoui authored
oups, correctly free the tcp clock/data when the component is freed, not only when the stream is manually removed through the API
-
Youness Alaoui authored
-
Youness Alaoui authored
-
Youness Alaoui authored
-
Youness Alaoui authored
Adding 'Since' gtk-doc tags to specify since which version each function has become available (for non 0.0.1 ones)
-
Youness Alaoui authored
-
Youness Alaoui authored
-
Youness Alaoui authored
-
Youness Alaoui authored
-
- 29 Dec, 2009 1 commit
-
-
Youness Alaoui authored
Keep the upnp and upnp-timeout properties available even if upnp is disabled during build. This is to prevent a break of the API from one build to another
-
- 13 Nov, 2009 1 commit
-
-
Youness Alaoui authored
use g_strlcpy instead of strncpy and memcpy for the candidate foundation, because it makes sure we have a trailing null char in our string.
-