Sjoerd Simons [Sun, 29 Jul 2012 16:31:37 +0000 (18:31 +0200)]
GST_FLOW_WRONG_STATE -> GST_FLOW_FLUSHING
Olivier Crête [Fri, 27 Jan 2012 17:55:09 +0000 (18:55 +0100)]
Use GstPushSrc as a base class
Olivier Crête [Fri, 27 Jan 2012 17:54:31 +0000 (18:54 +0100)]
Port GStreamer src/sink to 0.11
Livio Madaro [Fri, 3 Aug 2012 13:17:08 +0000 (15:17 +0200)]
Fix gaddr initialization
Olivier Crête [Wed, 6 Jun 2012 20:50:39 +0000 (16:50 -0400)]
Only define _FORTIFY_SOURCE if optimisation is enabled
Youness Alaoui [Mon, 16 Jul 2012 19:44:42 +0000 (15:44 -0400)]
Merge remote-tracking branch 'shadeslayer/dribble_mode'
Rohan Garg [Fri, 29 Jun 2012 19:33:34 +0000 (01:03 +0530)]
Also free the GCond and the GMutex
Rohan Garg [Thu, 21 Jun 2012 19:22:44 +0000 (00:52 +0530)]
Allocate memory to stun_mutex_ptr and stun_signal_ptr
Olivier Crête [Tue, 29 May 2012 01:02:31 +0000 (21:02 -0400)]
Cache GSocketAddress in UdpBsdSocket
Creating a GSocketAddress takes a measurable amount of time, so let's cache it
as it doesn't change once a socket is created.
Rohan Garg [Mon, 7 May 2012 18:31:48 +0000 (00:01 +0530)]
Multiple fixes for new test
* Make STUN thread drop STUN packets using a global variable
* Completely refactor the way the STUN thread exits
- Now exit by setting exit_stun_thread as TRUE and sending
dummy data
- Remove the old signal mechanism
* Don't set unnecessary socket options
* Lock stun_mutex_ptr before signalling STUN response
Rohan Garg [Thu, 3 May 2012 20:53:21 +0000 (02:23 +0530)]
Fixes to dribble mode test as per suggestions
- Use LEFT_AGENT and RIGHT_AGENT instead of calling GINT_TO_POINTER everytime
- Add padding (0's in our case) before the actual payload and make sure the
reciever checks for this padding. This ensures no STUN packets got through
- Free GSList's properly in cb_agent_new_candidate
- Use proper calls to g_signal_handlers_disconnect_by_func
Rohan Garg [Thu, 3 May 2012 17:47:20 +0000 (23:17 +0530)]
Minor bug fixes and enhancements to the test
Rohan Garg [Mon, 30 Apr 2012 21:50:49 +0000 (03:20 +0530)]
Update .gitignore for new test
Rohan Garg [Tue, 17 Apr 2012 16:04:00 +0000 (21:34 +0530)]
Use %lu for long unsigned int when calling printf
Rohan Garg [Wed, 4 Apr 2012 14:42:38 +0000 (20:12 +0530)]
Fix test-dribble
The ragent state can be either NICE_COMPONENT_CONNECTED or NICE_COMPONENT_READY
Rohan Garg [Mon, 2 Apr 2012 14:36:28 +0000 (20:06 +0530)]
Update documentation to reflect the dribble mode feature
Rohan Garg [Fri, 16 Mar 2012 19:22:00 +0000 (00:52 +0530)]
Added new tests for dribble mode in libnice
4 tests in total :
* Common to all :
- Gather lagent and ragent candidates
- Delay STUN response for lagent
- Make sure connchecks succeed and data is sent across
to lagent
* standard_test :
- Set remote candidates for lagent while lagent is still
gathering it's own candidates
* bad_credentials_test :
- Set incorrect credentials for both lagent and ragent
- Make sure connchecks fail
- Signal STUN response and set correct credentials
* bad_candidate_test :
- Set a bad remote candidate for lagent
- Make sure connchecks fail
- Signal STUN response and set correct remote candidates
* new_candidate_test :
- As soon as a new local candidate is found set it as
the remote candidate for the other agent
Rohan Garg [Sat, 10 Mar 2012 13:13:41 +0000 (18:43 +0530)]
Add conn_check_add_for_local_candidate to add the newly found local candidate to the conncheck list
conn_check_add_for_local_candidate works exactly like conn_check_add_for_candidate
except that it starts connchecks for newly found local candidates with existing
remote candidates
Refactor priv_add_local_candidate_pruned to accept a NiceAgent and a stream_id
in order to call conn_check_add_for_local_candidate for newly gathered local candidate
Rohan Garg [Sat, 10 Mar 2012 12:44:21 +0000 (18:14 +0530)]
Dribble Mode : Do not check whether we are still gathering our own candidates
Youness Alaoui [Tue, 3 Apr 2012 20:40:45 +0000 (16:40 -0400)]
version 0.1.2.1
Youness Alaoui [Tue, 3 Apr 2012 20:37:38 +0000 (16:37 -0400)]
version 0.1.2
Youness Alaoui [Tue, 3 Apr 2012 20:11:19 +0000 (16:11 -0400)]
Recognize a matching check if it's an INVALID message (error with no error-code)
Youness Alaoui [Tue, 3 Apr 2012 20:09:23 +0000 (16:09 -0400)]
Reinitiate a FAILED conncheck pair on a triggered check
Rohan Garg [Fri, 23 Mar 2012 08:02:23 +0000 (13:32 +0530)]
Use len instead of buf_len to check whether or not data was received properly
Rohan Garg [Sat, 10 Mar 2012 13:03:49 +0000 (18:33 +0530)]
Properly ifdef windows headers in order to get compiles working on linux again
Filippo Della Betta [Thu, 8 Mar 2012 21:19:02 +0000 (16:19 -0500)]
Added more tests on Visual Studio platform
Filippo Della Betta [Mon, 5 Mar 2012 16:55:59 +0000 (17:55 +0100)]
Fixed size of memcpy
Filippo Della Betta [Mon, 5 Mar 2012 17:07:08 +0000 (18:07 +0100)]
unistd.h is not present on Visual Studio. Replaced usleep with g_usleep to support win32 platform as well. Added WSAStartup/WSACleanup on win32
Filippo Della Betta [Mon, 5 Mar 2012 17:04:57 +0000 (18:04 +0100)]
Fixed mingw/msys compilation. Definition for intptr_t was missing. Added WSAStartup/WSACleanup on win32
Filippo Della Betta [Mon, 5 Mar 2012 17:04:12 +0000 (18:04 +0100)]
Fixed mingw/msys compilation. Definition for intptr_t was missing. Added WSAStartup/WSACleanup on win32
Filippo Della Betta [Mon, 5 Mar 2012 17:03:21 +0000 (18:03 +0100)]
Added WSAStartup/WSACleanup on win32
Filippo Della Betta [Mon, 5 Mar 2012 17:02:30 +0000 (18:02 +0100)]
Fixed mingw/msys compilation. Definition for intptr_t was missing
Filippo Della Betta [Mon, 5 Mar 2012 17:00:40 +0000 (18:00 +0100)]
Fixed compiling on Visual Studio and removed getsockname before bind. Added WSAStartup/WSACleanup on win32
Filippo Della Betta [Mon, 5 Mar 2012 16:57:53 +0000 (17:57 +0100)]
Fixed compile on mingw/msys platform
Filippo Della Betta [Mon, 5 Mar 2012 16:54:52 +0000 (17:54 +0100)]
Added win32 section to .gitignore
Filippo Della Betta [Mon, 5 Mar 2012 11:33:32 +0000 (12:33 +0100)]
Added support for compiling when gtk-doc is not installed
Filippo Della Betta [Mon, 5 Mar 2012 11:28:27 +0000 (12:28 +0100)]
Added gio dependency
Youness Alaoui [Fri, 2 Mar 2012 02:33:17 +0000 (21:33 -0500)]
Fix test-bsd by calling g_type_init
Youness Alaoui [Fri, 2 Mar 2012 02:33:06 +0000 (21:33 -0500)]
Fix tcp-bsd/udp-bsd coding style, add robustness and remove use of name_len
Youness Alaoui [Fri, 2 Mar 2012 02:31:52 +0000 (21:31 -0500)]
Fix memory leak with udp-bsd
Youness Alaoui [Fri, 2 Mar 2012 02:27:48 +0000 (21:27 -0500)]
Callback now takes a GSocket * as argument
Youness Alaoui [Fri, 2 Mar 2012 02:25:32 +0000 (21:25 -0500)]
Fix coding style
Livio Madaro [Wed, 29 Feb 2012 13:24:58 +0000 (14:24 +0100)]
Replace sockets with GSocket
Use GSocket for better Windows compatibility
Livio Madaro [Thu, 23 Feb 2012 15:39:08 +0000 (16:39 +0100)]
set nonblocking mode for socket on Windows
Filippo Della Betta [Thu, 23 Feb 2012 16:04:02 +0000 (17:04 +0100)]
Changed EOL to libnice.sln and added config.h
Filippo Della Betta [Mon, 20 Feb 2012 10:47:09 +0000 (11:47 +0100)]
Added Microsoft Visual Studio 2008 (v9.0) support
Rohan Garg [Sat, 18 Feb 2012 20:12:32 +0000 (01:42 +0530)]
Remove redundant code from tests
Filippo Della Betta [Wed, 15 Feb 2012 11:02:13 +0000 (12:02 +0100)]
WSAStartup/WSACleanup added on WIN32 platform
Filippo Della Betta [Fri, 17 Feb 2012 10:49:26 +0000 (11:49 +0100)]
Fixed warnings on Visual Studio platform
Filippo Della Betta [Fri, 17 Feb 2012 10:47:35 +0000 (11:47 +0100)]
Added fix for Visual Studio platform
Filippo Della Betta [Wed, 15 Feb 2012 11:02:13 +0000 (12:02 +0100)]
Replaced __func__ macro with G_STRFUNC in the unit tests
Youness Alaoui [Wed, 15 Feb 2012 21:13:31 +0000 (16:13 -0500)]
Fix peer reflexive valid pair priority.
Should use the remote priority rather than the priority of the parent pair
Thanks to Filippo Della Betta
Filippo Della Betta [Wed, 15 Feb 2012 11:03:28 +0000 (12:03 +0100)]
fixed warning guint32/guint64
Filippo Della Betta [Wed, 15 Feb 2012 11:00:54 +0000 (12:00 +0100)]
Error definitions moved to pseudotcp.h on WIN32 platform
Filippo Della Betta [Wed, 15 Feb 2012 10:13:18 +0000 (11:13 +0100)]
Fix declaration after statement error on VS
Since Visual Studio compiler does not support C99 I put some declarations
at the beginning of the statement.
Filippo Della Betta [Wed, 15 Feb 2012 10:07:57 +0000 (11:07 +0100)]
Added typedef on size_t and ssize_t that are not defined on Visual Studio platform
Filippo Della Betta [Wed, 15 Feb 2012 10:04:39 +0000 (11:04 +0100)]
Included the replacement header win32_common.h instead of the missing header stdint.h on WIN32 platform
Filippo Della Betta [Wed, 15 Feb 2012 09:58:58 +0000 (10:58 +0100)]
Use g_io_channel_win32_new_socket for windows
As glib documentation suggests (http://developer.gnome.org/glib/2.28/glib-IO-Channels.html),
g_io_channel_unix_new should be avoided on WIN32 platform.
Replaced with g_io_channel_win32_new_socket function
Rohan Garg [Tue, 14 Feb 2012 17:24:17 +0000 (22:54 +0530)]
Use size of the MD5Context, not size of the pointer
Rohan Garg [Tue, 14 Feb 2012 17:19:50 +0000 (22:49 +0530)]
Use the correct enum
Youness Alaoui [Mon, 13 Feb 2012 20:12:36 +0000 (15:12 -0500)]
Return real origin when using a proxy (fixes detecting TURN sockets)
If using a proxy, the 'from' of all packets were of the proxy server
instead of the turn server, causing the check for turn to fail.
Thanks to Madaro Livio.
Madaro Livio [Mon, 13 Feb 2012 20:03:00 +0000 (15:03 -0500)]
udp-bsd: Do not return an error for CONNRESET errors from ICMP
I have a use case where len < 0 but the socket is still working.
It is when the stun server is not listening and the client receives
ICMP messages. I think it is a Windows specific problem.
Calling WSAGetLastError() returns WSAECONNRESET.
The MSDN documentation sais:
"On a UDP-datagram socket, this error would indicate that a previous
send operation resulted in an ICMP "Port Unreachable" message."
http://msdn.microsoft.com/en-us/library/windows/desktop/ms740121(v=vs.85).aspx
Madaro Livio [Fri, 10 Feb 2012 23:45:19 +0000 (18:45 -0500)]
Do not crash if the TURN server is invalid or not listening
If TURN server is not listening then nice_tcp_bsd_socket_new
returns socket=NULL and libnice crashes.
Youness Alaoui [Thu, 9 Feb 2012 23:53:27 +0000 (18:53 -0500)]
oups, version should be 0.1.1.1
Youness Alaoui [Thu, 9 Feb 2012 23:45:04 +0000 (18:45 -0500)]
stund: Use recvfrom/sendto instead of recvmsg/sendmsg
John Selbie [Thu, 9 Feb 2012 23:40:19 +0000 (18:40 -0500)]
stund: Set message length (fixes infinite loop) when receiving unknown attributes
Youness Alaoui [Thu, 9 Feb 2012 03:46:07 +0000 (22:46 -0500)]
Merge branch 'master' of ssh://dhansak/srv/git.collabora.co.uk/git/libnice
Conflicts:
configure.ac
Madaro Livio [Thu, 9 Feb 2012 03:17:25 +0000 (22:17 -0500)]
Keep TURN credentials (realm/nonce) from allocation for the refresh request.
Calling nice_agent_remove_stream will try to release the TURN allocation (lifetime=0)
but it won't work because the CandidateRefresh doesn't contain credentials
for user authentication (if it happens right after the allocation).
The server replies with an authentication error but the client closed the socket
and does not reply to it.
Madaro Livio [Thu, 9 Feb 2012 03:03:14 +0000 (22:03 -0500)]
The server reflexive candidate gathered from the turn server is
not valid if the socket is reliable (TCP).
The patch does not add the candidate to the local candidate list
if the socket is reliable.
Madaro Livio [Thu, 9 Feb 2012 02:59:31 +0000 (21:59 -0500)]
Disable retransmission of turn messages for reliable socket.
If a client connects to a TURN server using a reliable socket (TCP)
then retransmission of TURN messages (allocation request) is not needed.
Libnice 0.1.1 retransmit the TURN messages (allocation request) in the
same way for reliable (TCP) and not reliable (UDP) socket.
Madaro Livio [Thu, 9 Feb 2012 02:56:45 +0000 (21:56 -0500)]
Fix sending data padding for TURN TCP draft 9 and RFC 5766
Data padding is required for interoperability with standard TURN server.
Youness Alaoui [Fri, 3 Feb 2012 01:29:58 +0000 (20:29 -0500)]
only create the unknown-attributes error if we are going to send it
Youness Alaoui [Fri, 3 Feb 2012 01:29:35 +0000 (20:29 -0500)]
some debug message
Youness Alaoui [Fri, 3 Feb 2012 01:29:13 +0000 (20:29 -0500)]
Mark pair as nominated when it's a remote peer-reflexive candidate discovered through an early incoming check which has use-candidate
Youness Alaoui [Thu, 2 Feb 2012 21:29:01 +0000 (16:29 -0500)]
Refuse remote candidates on a stream only if that specific stream is gathering.
This should fix the use case of creating a stream 1, gather until finished
then creating another stream, gather its candidates, then setting the
remotes on stream 1 (which is already gathered) while stream 2 is still gathering.
Youness Alaoui [Thu, 2 Feb 2012 21:32:04 +0000 (16:32 -0500)]
Fix missing semicolon from previous commit in test-restart
Rohan Garg [Mon, 16 Jan 2012 22:36:03 +0000 (04:06 +0530)]
Fix builds for systems using glib 2.3.81 and above
Youness Alaoui [Fri, 13 Jan 2012 21:13:52 +0000 (16:13 -0500)]
Do not check for upnp_mapping if UPnP is disabled at compile time
Youness Alaoui [Thu, 12 Jan 2012 23:27:42 +0000 (18:27 -0500)]
test-fullmode: if upnp is disabled, then disable it in both agents
Youness Alaoui [Thu, 12 Jan 2012 23:27:20 +0000 (18:27 -0500)]
Fix typo: Cancel the UPnP timeout if all scheduled upnp requests are done
Youness Alaoui [Thu, 12 Jan 2012 23:26:49 +0000 (18:26 -0500)]
Do not signal gathering-done if UPnP is still waiting for an answer
Siraj Razick [Tue, 3 Jan 2012 22:10:21 +0000 (17:10 -0500)]
Add -no-undefined to ldflags
This ensures that all the symbols are defined during link time, and also
without this option .dll files for windows are not produced.
Siraj Razick [Tue, 20 Dec 2011 19:00:04 +0000 (14:00 -0500)]
Fixes the linker errors we get under mingw32
When we link agaist libnice under ming32, the client applications
fails to link. this patch fixes that
Siraj Razick [Mon, 19 Dec 2011 22:18:43 +0000 (17:18 -0500)]
Changes windows specific headers to lower case
This helps us when cross compiling libnice for Windows from Linux
Youness Alaoui [Mon, 12 Dec 2011 23:50:34 +0000 (18:50 -0500)]
Set HAVE_GETIFADDRS only if ifaddrs.h is found, not if any of arpa/inet.h net/in.h and ifaddrs.h are found
Nader Morshed [Wed, 30 Nov 2011 20:36:51 +0000 (15:36 -0500)]
Add support for windows's compilation through MinGW
Youness Alaoui [Mon, 28 Nov 2011 13:08:44 +0000 (08:08 -0500)]
Update TODO
Youness Alaoui [Fri, 4 Nov 2011 14:57:34 +0000 (10:57 -0400)]
Change glib dependency to >= 2.13 because of G_PARAM_STATIC_STRINGS
Timothy Redaelli [Thu, 13 Oct 2011 00:08:14 +0000 (02:08 +0200)]
Update .gitignore
Timothy Redaelli [Thu, 13 Oct 2011 00:08:13 +0000 (02:08 +0200)]
Remove warnings on stun/tests
Timothy Redaelli [Thu, 13 Oct 2011 00:08:12 +0000 (02:08 +0200)]
Use only POSIX statements in nice/Makefile.am
Olivier Crête [Wed, 21 Sep 2011 03:33:39 +0000 (23:33 -0400)]
Don't pass the agent to the socket layer
Olivier Crête [Wed, 21 Sep 2011 03:08:21 +0000 (23:08 -0400)]
Keep a ref on the GMainContext that are attached to
Olivier Crête [Wed, 21 Sep 2011 02:22:10 +0000 (22:22 -0400)]
Use the agent's main main-context for the tcp-bsd callback as it's for sending
Olivier Crête [Wed, 21 Sep 2011 03:51:24 +0000 (23:51 -0400)]
Version 0.1.1.1
Youness Alaoui [Thu, 8 Sep 2011 02:01:18 +0000 (22:01 -0400)]
version 0.1.1.0
Youness Alaoui [Thu, 8 Sep 2011 00:35:01 +0000 (20:35 -0400)]
version 0.1.1
Apostolos Syropoulos [Thu, 8 Sep 2011 00:59:03 +0000 (20:59 -0400)]
Fix support for solaris
Olivier Crête [Tue, 23 Aug 2011 23:19:09 +0000 (19:19 -0400)]
Google puts the STUN cookie, but doesn't follow RFC 5389
Google puts the stun cookie, but it doesn'T follow RFC 5389 at all,
so ignore the STUN cookie in Google mode.
David Woodhouse [Mon, 15 Aug 2011 19:30:36 +0000 (15:30 -0400)]
Fix check-symbols test for PPC64