August 2013 - freeswitch-dev@lists.freeswitch.org

[Freeswitch-dev] accessing key presses with outbound event socket
by Jamie McClelland 02 Sep '13

02 Sep '13

Hi all, Thanks again for your help in completing our simulataneous interpretation system using freeswitch[0]. I'm now trying to add the ability for users to enter the conference number they would like to join. I see lots of examples for doing this, but none that are using an outbound event socket. With event socket, I can capture dtmf digit events *after* the user enters the conference call. But I am having trouble first prompting the user to enter the conference number and then capturing their output and using it to determine which conference room to sent them to. I'm using fiorix's python event socket code [1]. So far I've been experimenting with the read command[2]. However, I'm not sure how to retrieve the digits pressed after calling this command. Is their a command that can query freeswitch to get the value of the variable I set with the read command? Any help would be greatly appreciated. jamie 0. http://lists.freeswitch.org/pipermail/freeswitch-dev/2013-February/006406.h… 1. https://github.com/fiorix/eventsocket 2. http://wiki.freeswitch.org/wiki/Misc._Dialplan_Tools_read -- May First/People Link Growing networks to build a just world http://www.mayfirst.org https://support.mayfirst.org OpenPGP Key: http://current.workingdirectory.net/pages/identity/

2 3

[Freeswitch-dev] Event dispatch overview
by Ola Bäckström 28 Aug '13

28 Aug '13

Hi all, I've been trying to control eavesdrop with events the last couple of days, but not succeeded. Since I needed to know more how events are dispatched internally to the different channels I've been diving into Freeswitch source code to get some insight. I thought I could share what I found, and hopefully get some comments or correction. Especially on what I did not find out ... in particular I haven't understood where/how the event dispatch to channels work, I call it "Magic" below... ** Overview of the Freeswitch event system (see the file src/include/switch_event.h for an overview in the code) 1. At the core there is an event dispatch queue. (see switch_events.c) 2. Subscribers can request to be notified when specific events are handled. All mod_xxx stuff does this, see mod_erlang_event, mod_dptools etc (see switch_event_bind(...) function, used to register a handler function for specific events) 3. Events can be created anywhere in any thread (see switch_event_create*-functions) and edited, by adding key-value pairs called headers (see switch_event_add_header(...) function). Example of this can be found in mod_event_test.c function torture_thread(...). 4. Such an "unsent" event can be put into the event dispatch queue (see switch_event_fire(...), that function calls switch_event_queue_dispatch(...) that finally calls switch_queue_push(...) to push it onto the queue) 5. Later... will all relevant subscribers will get their callback handler function called with the event. **Event handling inside a channel Each channel contains a couple of queues: event_queue, message_queue, signal_data_queue, private_event_queue, private_event_queue_pri. (see the type switch_core_session_t that equals the struct switch_core_session found in include/private/switch_core_pvt.h) To make sure an event ends up in one specific channel queue the correct headers must be added between step 3 and 4 above. (see switch_channel_event_set_data(...) ). Example of this can be found in switch_ivr_async.c function switch_ivr_broadcast(...), or in mod_dptools.c where all dialplan commands are implemented. Then by some unclear mechanism "Magic" will the event leave the central event dispatch queue and end up in one of the channel queues. **Core State machine Unclear what role it has, but switch_core_state_machine.c has a mechanism to call driver state handlers depending on the channel state. Events does not seem to be directly involved in those calls. Regards /Ola

1 0

[Freeswitch-dev] Fixed bug
by Oleg Khovayko 27 Aug '13

27 Aug '13

Hi, I built fresh FreeSWITCH from git. OS FreeBSD 9.1. Got compilation error: cc1: warnings being treated as errors tport_type_ws.c: In function 'tport_recv_stream_ws': tport_type_ws.c:231: warning: format '%ld' expects type 'long int', but argument 11 has type 'ssize_t' gmake[9]: *** [tport_type_ws.lo] Error 1 gmake[8]: *** [all] Error 2 gmake[7]: *** [all-recursive] Error 1 gmake[6]: *** [all-recursive] Error 1 gmake[5]: *** [all] Error 2 gmake[4]: *** [/usr/src/freeswitch-upstream/libs/sofia-sip/libsofia-sip-ua/libsofia-sip-ua.la] Error 2 gmake[3]: *** [mod_sofia-install] Error 1 gmake[2]: *** [install-recursive] Error 1 gmake[1]: *** [install-recursive] Error 1 gmake: *** [install] Error 2 [root@olegh /usr/src/freeswitch-upstream]# Found and fixed wrong conversion in the file (see bold (long) below): ./libs/sofia-sip/libsofia-sip-ua/tport/tport_type_ws.c 231 SU_DEBUG_1(("%s(%p): su_getmsgsize(): %s (%d) N=%ld\n", __func__, (void *)self, 232 su_strerror(err), err, *(long)*N)); Please, forward to appropriate maintainer. Thanks, Oleg

2 1

[Freeswitch-dev] Authoring Modules, moving mymodule.conf.xml to autoload config
by Joel Dodson 27 Aug '13

27 Aug '13

Hi, I'm writing a new module for some specific functionality we need. I've been going through the wiki page for Authoring Modules (http://wiki.freeswitch.org/wiki/Authoring_Freeswitch_Modules) and it's been helpful. I have mod_mymodule (copied from mod_skel) compiling and running locally. I've also been looking through mod_event_socket since my module will be very similar. I'm stuck now on getting the config file I've included in my module to be automagically added to the autoload_configs directory. I've added a conf directory to my module similarly to how event_socket does it. I've built my module (make mod_mymodule and make mod_mymodule-install) but mymodule.conf.xml still is not moved automatically into the autoload directory. I do have mod_mymodule included in modules.conf when I build and install. I've even tried running bootstrap.sh and configure again with my mod_mymodule in the source code. Do I need to manually copy over mymodule.conf.xml when I do a freeswitch install? Is there another file I need to add mod_mymodule to to get the config copied over automatically? I'm doing this on the latest 1.2 branch from github, I've tried it on master as well, same results. thanks, Joel

2 1

[Freeswitch-dev] Problem send message from code
by Juan Antonio Ibañez Santorum 27 Aug '13

27 Aug '13

Hello, I have the following code to send SIP MESSAGE to some endpoins. The problem is that sometimes works, sometimes not and I am not be able to find why. Could you find any problem in that function? Is there any other way to send messages? http://pastebin.com/fnQj00Eg Regards

2 3

[Freeswitch-dev] Thanks to all those that make ClueCon Possible!!
by Ken Rice 26 Aug '13

26 Aug '13

ClueCon 2013 was a smashing success and you had everything to do with it! We were very glad to have you with us this year and we hope you enjoyed it every bit as much as we did. The ClueCon team is currently working on getting all of the videos and presentation slides put up on the ClueCon website. More information will be sent out soon. In the meantime be sure to join us each Wednesday at 1PM Eastern time for the ClueCon Weekly conference call. What about ClueCon 2014? It's already in the works! 2014 marks the 10th anniversary of ClueCon - a milestone you'll not want to miss. We will be putting out a call for speakers as well as getting our sponsors arranged in the next few months. Stay tuned for more information. If you missed Michael Giagnocavo's ClueCon 2013 presentation, "Easy Real-Time CDR Reporting with VoltDB," check it out here <http://goo.gl/EywbDm> . And then grab a free trial of VoltDB <http://go.voltdb.com/voltdb-enterprise-software-download> the database that helps Michael manage high-volume CDRs. The ClueCon team would like to thank the following sponsors: Nenad, Moises, and all the good people at Sangoma <http://www.sangoma.com> . Sangoma has been with us since the very first ClueCon. Sean, Jeff, and the folks at Flowroute <http://www.flowroute.com> . Arnie and the gang at NACT <http://www.nact.com> . Kashif and the team over at Vestec <http://www.vestec.com> . Norm and the VoiceNetwork <http://www.voicenetwork.ca> team. Venky and the good folks at Plivo <http://www.plivo.com> . The whole OrecX <http://www.orecx.com> team. John, Nicole and the rest of the gang at OnSIP <http://www.onsip.com> . Darren and all the techies at 2600hz <http://www.2600hz.com> . John, Dave, and all the rest of the Truphone <http://www.truphone.com> guys. Kevin and the team at Ooma <http://www.ooma.com> . Tommy and the folks at Snom <http://www.snom.com> . The tokbox <http://www.tokbox.com> team. The NG Communications <http://www.ngcommunications.com> team. We hope to see you all again in 2014! -- Ken http://www.FreeSWITCH.org http://www.ClueCon.com http://www.OSTAG.org irc.freenode.net #freeswitch

1 0

[Freeswitch-dev] Patchs Jira and more....
by Ken Rice 21 Aug '13

21 Aug '13

Hey Guys, there seems to be a rash of people uploading patchs to Jira lately that just do not work... So lets hit this one really quick... 1. if you are patching a generated File like the Makefile or configure script make sure you are patching the files they are generated from... ie: configure.in for configure... 2. don¹t try to diff 2 separate trees... git diff > /tmp/my.fancy.patch.txt¹ from the root of your fs source checkout will give you a plain text file of the patch suitable for uploading to jira. 3. Please include all information you have on the jira and what problem/feature the patch addresses. Don¹t just say ³here¹s a patch² and expect someone to look at it without knowing what it is. 4. If a commiter comments on your jira that you need to freshen the patch please do so in a timely fashion or your patch will go back to the bottom of the queue as they will move on to something else... This isnt ignoring your efforts, but developers tend to get distracted easily with new shiny objects so getting back to them quickly helps tremendously. These simple things will help keep the devs from going crazy! Now for you guys that use git on a regular basis, before you blow a gasket, yes this is very basic, and is intended to be so... Yes there are other (and possibly arguable better ways to generate a patch, but this way is the simplest for people not familiar with git) -- Ken http://www.FreeSWITCH.org http://www.ClueCon.com http://www.OSTAG.org irc.freenode.net #freeswitch

1 0

[Freeswitch-dev] Keepalive Value for TCP/TLS sockets in Freeswitch
by Master Can 20 Aug '13

20 Aug '13

Hello, I noticed that when I do a "netstat --timers | grep sip-tls" I see: keepalive (2,62/0/0) The first value counts from 30 seconds downwards. So the freeswitch TCP connection to my client seems to have a keepalive value of 30 secs. I want to get rid of this since my clients send keepalive packets themselves every x seconds (currently every 600 seconds). How can I increase or disable the Freeswitch keepalive function? I first thought that I'd need to set: <param name="nat-options-ping" value="false"/> <param name="all-reg-options-ping" value="false"/> but that didn't help. (My client is behind a NAT, and I thought this 30 seconds thing might be due to OPTIONS packets being sent every 30 seconds...) Are there any suggestions how to turn off the keepalive function? br, Can

1 0

[Freeswitch-dev] Freeswitch signaling issue when bridge call from FS to asterisk server.
by ThanhTruong 20 Aug '13

20 Aug '13

Hi all, I have a small issue with my testing system with asterisk and FS server. I have 2 servers. One is Fs and one is asterisk server. I open incoming call from asterisk server(for all call coming). In Fs server, I have routed call in public.xml like: <extension name="ROUTE"> <condition field="destination_number" expression="(\d+)$" require-nested="true"> <condition field="network_addr" expression="^X\.X\.X\.X$" break="never"> <action application="set" data="hangup_after_bridge=true"/> <action application="export" data="sip_h_X-ROUTE=LOOKUP"/> <action application="bridge" data="user/${destination_number}@${domain_name}"/> <anti-action application="set" data="hangup_after_bridge=true"/> <anti-action application="export" data="sip_h_X-ROUTE=LOOKUP"/> <anti-action application="bridge" data="sofia/internal/${destination_number}(a)X.X.X.X"/> </condition> </condition> </extension> I have 2 client : + Xlite registered in Fs server + Eyebeam registered in asterisk server. Then I have tried some several case and get some results as: + when I make call from xlite to eyebeam , call connected and: - When i hangup from xlite, eyebeam keep connected and not hangup - when I hangup from eyebeam, xlite keep connected and not hangup too. +Then, I have tried to make call from eyebeam to xlite, call connected and: - I hangup call in xlite, eyebeam hangup. - I hangup call in eyebeam, xlite hangup. -----------------------> only can hangup in this case. + Then I have tried another case: I have call from eyebeam to xlite and vise versace : call auto hangup after 30 seconds. I have thought that is NAT issue, But If I register direct xlite and eyebeam to asterisk server, call will ok, and there is not any issue there. So, Hope some one can suggest some thing to fix it. Thanks, ThanhTruong

3 3

[Freeswitch-dev] Issue when receiving INVITE with both audio and image (T38)
by François 16 Aug '13

16 Aug '13

Hello list, I've seen several gateways (Patton, Audiocodes, Nortel CS2K), sending both audio+image (T38) in the initial INVITE. In a simple bridge scenario, FS will just ignore the T38 part and setup an audio connection only with b-leg. My issue is later on if/when we receive a re-INVITE (ex: SIP timers refresh) with the same SDP (audio+T38), FS now considers it as FAX and sends a re-INVITE to b-leg with T38 only! In case the destination is voice only, the call hangs up. --- Asterisk apparently had the same issue, they solved it by setting up both streams on a-leg (RTP + UDPTL). If they receive packets on UDPTL they just re-INVITE b-leg with T38. See: https://reviewboard.asterisk.org/r/208/ Also: https://issues.asterisk.org/jira/browse/ASTERISK-11843 Is this the right way to go? (Also is it feasible in FreeSwitch)? I can add a JIRA if necessary. Thanks, François.

1 0