iaxclient-devel — Development Discussion List

Re: [Iaxclient-devel] contrib/tcl music on hold patch

[Andrea S. <si...@op...>]

Hi Mats

Mats Bengtsson wrote:
> Andrea,
> 
> Perhaps you have any tips for building my tcl package on linux.

hopefully ;)

> I have made a memo of my procedures in
> contrib/tcl/build-iaxclient.txt
> but when I try to load it into tcl it complains about missing speex symbol
> speex_preprocess_ctl
> Also, it is just about 280k which seems way too small, so I suspect some
> missing code there.

I've compiled required libs a month ago
so take the following with a grain of salt

regarding lib speex that's what I've done so far:

download speex 1.2beta3  (http://downloads.xiph.org/releases/speex/speex-1.2beta3.tar.gz)
./configure && make && sudo make install

as you can see I ran ./configure with default values
and the default --prefix value is /usr/local
so you've everything installed in /usr/local/lib

that's what I've got

sickpig@suino:~$ls -lhct /usr/local/lib/ | grep speex
-rw-r--r-- 1 root root  221K 2008-02-18 10:26 libspeexdsp.a
-rwxr-xr-x 1 root root   840 2008-02-18 10:26 libspeexdsp.la
lrwxrwxrwx 1 root root    20 2008-02-18 10:26 libspeexdsp.so -> libspeexdsp.so.1.4.0
lrwxrwxrwx 1 root root    20 2008-02-18 10:26 libspeexdsp.so.1 -> libspeexdsp.so.1.4.0
-rwxr-xr-x 1 root root  181K 2008-02-18 10:26 libspeexdsp.so.1.4.0
-rw-r--r-- 1 root root  336K 2008-02-18 10:26 libspeex.a
-rwxr-xr-x 1 root root   819 2008-02-18 10:26 libspeex.la
lrwxrwxrwx 1 root root    17 2008-02-18 10:26 libspeex.so -> libspeex.so.1.4.0
lrwxrwxrwx 1 root root    17 2008-02-18 10:26 libspeex.so.1 -> libspeex.so.1.4.0
-rwxr-xr-x 1 root root  244K 2008-02-18 10:26 libspeex.so.1.4.0


whereas I've built iaxclient lib using this configure option:

./configure --disable-video --enable-clients=testcall --with-local-gsm
make
sudo make install

result:

sickpig@suino:~$ls -lcht  /usr/lib/tcliaxclient0.2/
total 48K
-rw-r--r-- 1 root root 5.9K 2008-04-17 10:28 iaxclient.tcl
-rw-r--r-- 1 root root  185 2008-04-17 10:28 pkgIndex.tcl
-rwxr-xr-x 1 root root  34K 2008-04-17 10:28 libtcliaxclient0.2.so



Andrea

Re: [Iaxclient-devel] contrib/tcl music on hold patch

[Mats B. <ma...@gm...>]

Andrea,

Perhaps you have any tips for building my tcl package on linux.
I have made a memo of my procedures in
contrib/tcl/build-iaxclient.txt
but when I try to load it into tcl it complains about missing speex symbol
speex_preprocess_ctl
Also, it is just about 280k which seems way too small, so I suspect some
missing code there.

When I run make the link command is:

gcc -pipe -shared -o libtcliaxclient0.2.so iaxclient.o tones.o
XThreadUtil.o -lportaudio -lspeexdsp -lspeex -liaxclient -lpthread
-lasound  -L/usr/local/lib -ltclstub8.5

and I don't have any libspeex.so etc. installed on my box as far as I can see.

Mats

Re: [Iaxclient-devel] v2.1beta3 need patch ?

[Peter G. <jpg...@gm...>]

Hi Jean-François,

On Wed, Apr 16, 2008 at 2:18 PM, Jean-François Guchens
<jf....@sq...> wrote:
> I think i've found a bug in iaxclient_lib.c :
>  Line 1168 need to be commented. (calls[callNo].state |=
>  IAXC_CALL_STATE_COMPLETE;)

This behavior has been in iaxclient for a very long time. I will need
more information to be convinced that the semantic here is wrong.

>  Else when receiving an incoming call the call state is always marked as
>  complete.

Could you describe the exact scenario? I'm not sure I understand what
the problem is. It seems like this may be related to what Ruslan
reported earlier this week -- have you read that thread?

Does this affect calls directly between iaxclient applications?

>  I patched my local version and it works fine now.

What about for calls going through asterisk?

It seems like if you made an outbound call with iaxclient this change
would prevent the call from ever getting to the completed state.

Pete

[Iaxclient-devel] C# IAXClient Wrapper

[Chris H. <ch...@as...>]

Asteria Solutions Group is pleased to announce the first release of the
C# IAXClient Wrapper. for vs.NET 2005.

This code is based on the Visual Basic.NET IAXClient Wrapper developed
by Andrew Pollack at Second Signal.  

For more information see:

http://www.asteriasgi.com/?q=IAXClient-Wrapper

[Iaxclient-devel] v2.1beta3 need patch ?

[Jean-François G. <jf....@sq...>]

I think i've found a bug in iaxclient_lib.c :
Line 1168 need to be commented. (calls[callNo].state |= 
IAXC_CALL_STATE_COMPLETE;)
Else when receiving an incoming call the call state is always marked as 
complete.

I patched my local version and it works fine now.


jf

Re: [Iaxclient-devel] iaxClient patch to meet radio keying specs

[Peter G. <jpg...@gm...>]

Hi Steven

On Wed, Apr 16, 2008 at 9:02 AM, Steven Henke <sp...@xe...> wrote:
> Hello Project Leaders,
>
>  Please help me with the process for this project.
>
>  I modified my local working copy of iaxClient so that it provides the
>  radio tranceiver functions described in the IAX2 specifications.
>
>  I updated my working copy of the sources and produced the attached patch
>  file.
>
>  The resulting iaxclient library has been tested using the iaxRpt radio
>  dispatch client.
>
>  I would like this patch to be applied to the trunk so that it supports
>  continuing work with radio control applications.

I'm not too familiar with the radio part of the iax2 spec, but the
patch looks safe and the magic numbers seem to match those in
asterisk. I don't see any reason to not include the patch in
iaxclient.

The big question is whether we should squeeze this into the 2.1
release or just put it on trunk where it will have to wait until the
next release. I don't know when the next release will happen. IMHO,
this is safe enough to put into 2.1.

Objections?

>  Let me know if there is anything else I need to do to accomplish this.

The patch seems sufficient. Thank you for the contribution.

Pete

Re: [Iaxclient-devel] remote data is wrong

[Dennis C. <DCh...@pi...>]

On 15-Apr-08, at 11:59 AM, Peter Grayson wrote:

> On Tue, Apr 15, 2008 at 11:32 AM, Dennis Christopher
> <DCh...@pi...> wrote:
>> Hi Peter,
>>
>>  (The data size is reported as 320, as I emailed. The server's  
>> format is
>> ulaw instead of speex, if that matters.)
>
> I do not have a lot of experience with ulaw, but as I recall it also
> uses 20ms packets.
>
>>  Is there any further direction to look here? has anyone else got  
>> the raw
>> data to dump out properly
>>  on OSX?
>
> You could add some print statements to iaxclient to help understand
> what is going on. Search the code for
> "IAXC_AUDIO_PREF_RECV_REMOTE_RAW" -- those are good starting points.
>
> Pete

Pete,

Silly mistake on my part. All is working well with recording of  
remote data
via setting IAXC_AUDIO_PREF_RECV_REMOTE_RAW in the prefs. If
anyone in the future is looking for making it work under OS X I can  
provide
an example. Thanks for your help.

Dennis

Re: [Iaxclient-devel] contrib/tcl music on hold patch

[Andrea S. <si...@op...>]

Mats Bengtsson wrote:
> Many Thanks for this.

you're welcome

> I don't think the hold/unhold commands got tested at all.
> Now committed to svn.

perfect

Andrea

Re: [Iaxclient-devel] contrib/tcl music on hold patch

[Mats B. <ma...@gm...>]

Many Thanks for this.
I don't think the hold/unhold commands got tested at all.
Now committed to svn.

/Mats

On Wed, Apr 16, 2008 at 3:23 PM, Andrea Suisani <si...@op...> wrote:
> Hi all,
>
>  I was playing with music on hold through tcl extension
>  placed in contrib/tcl, I was able to successfully hold a
>  call but while trying to unhold my application simply
>  die (segfault).
>

[Iaxclient-devel] contrib/tcl music on hold patch

[Andrea S. <si...@op...>]

Hi all,

I was playing with music on hold through tcl extension
placed in contrib/tcl, I was able to successfully hold a
call but while trying to unhold my application simply
die (segfault).

Fortunately a similar case came to my mind while debugging
that issue, see this thread to get an idea:

http://sourceforge.net/mailarchive/message.php?msg_name=472846D8.3010405%40opinioni.net

After applying the same cure to contrib/tcl/iaxclient.c
I was able to unhold a call without having my app segfaulting.


that's the patch against current trunk

sickpig@suino:~/src/iaxclient-svn-tree/trunk/contrib/tcl$svn diff -x '-bwu' iaxclient.c
Index: iaxclient.c
===================================================================
--- iaxclient.c (revision 1412)
+++ iaxclient.c (working copy)
@@ -696,7 +696,7 @@
      }
      if ( result == TCL_OK ) {
         iaxc_quelch(selected, 1);
-       iaxc_select_call(-1);
+       //iaxc_select_call(-1);
      }
      return result;
  }
@@ -1089,7 +1089,7 @@
      }
      if (result == TCL_OK) {
         iaxc_unquelch(selected);
-       iaxc_select_call(selected);
+       //iaxc_select_call(selected);
      }
      return result;
  }


Andrea

[Iaxclient-devel] iaxClient patch to meet radio keying specs

[Steven H. <sp...@xe...>]

Hello Project Leaders,

Please help me with the process for this project.

I modified my local working copy of iaxClient so that it provides the
radio tranceiver functions described in the IAX2 specifications.

I updated my working copy of the sources and produced the attached patch
file.

The resulting iaxclient library has been tested using the iaxRpt radio
dispatch client.

I would like this patch to be applied to the trunk so that it supports
continuing work with radio control applications.

Let me know if there is anything else I need to do to accomplish this.

Thanks,

Steve Henke, W9SH

Re: [Iaxclient-devel] remote data is wrong

[Peter G. <jpg...@gm...>]

On Tue, Apr 15, 2008 at 11:32 AM, Dennis Christopher
<DCh...@pi...> wrote:
> Hi Peter,
>
>  (The data size is reported as 320, as I emailed. The server's format is
> ulaw instead of speex, if that matters.)

I do not have a lot of experience with ulaw, but as I recall it also
uses 20ms packets.

>  Is there any further direction to look here? has anyone else got the raw
> data to dump out properly
>  on OSX?

You could add some print statements to iaxclient to help understand
what is going on. Search the code for
"IAXC_AUDIO_PREF_RECV_REMOTE_RAW" -- those are good starting points.

Pete

Re: [Iaxclient-devel] Iaxclient Browser Plugin

[Andrea S. <si...@op...>]

Cristian Sepulveda wrote:
> Hi Everyone,
> 
> We are developing RuMPI, a plugin for Browser to enable realtime
> interaction, starting for VoIP using IAXClient, and later it will have
> video&P2P.

very interesting !

> This is not like SoftPhone, because all parameter are passed from server, so
> you could control it from the app. (where to register and where to call), we
> think  is perfect to enable click to call services using Asterisk.
> If you are interested to develop or test it, feel free to contact me.

I will do, I'm subscribing right now to rumpi-devel ml :)

> http://rumpi.sourceforge.net/

which is the actual stage of the project?


> PS: Next week I'll go to web2expo.com and I'll stay for a Month in SFO,
> could be really nice to meet face-to-face one of you and talk about
> IaxClient, Asterix and other stuff.

I will be more than happy to meet you there
but unfortunately I will be in the old Europe
at least foe the next few months :)

andrea

Re: [Iaxclient-devel] Problem with iaxc_register

[Jean-François G. <jf....@sq...>]

Sorry,
i was misunderstanding that iaxc_register is in fact just like a 
"register => foo:bar@server" in iax.conf.

regards

Jean-François Guchens a écrit :
> Hi,
>
> I don't understand how to register to an iax server an place a call.
> For exemple trying with testcall program gived with sources.
>
> Supposing i've a peer in my iax.conf :
>
> [jf]
> type=friend
> secret=foo
> auth=plaintext
> host=dynamic
> context=internal
> peercontext=internal
> qualify=yes
>
>
> and in my extensions.conf
>
> [internal]
> exten => 888,1,Dial(IAX2/888@10.0.0.4)
>
> When trying with testcall :
>
> ./testcall -u jf -p foo -h 10.0.0.2 888
>
> OR :
> ./testcall -u jf -p foo -h 10.0.0.2 888@internal
>
> i see in my cli :
> : chan_iax2.c:6027 update_registry: Restricting registration for peer 
> 'jf' to 60 seconds (requested 300)
>
> But no call.. no sound.. nothing happen.
>
> If i try :
> ./testcall -u jf -p foo -h 10.0.0.2 888@10.0.0.4
>
> It runs very well...
>
> I don't understand why i can't use exten of my context and how to avoid 
> this strange behavior.
>
> regards,
>
> jfg
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
> Don't miss this year's exciting event. There's still time to save $100. 
> Use priority code J8TL2D2. 
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
> _______________________________________________
> Iaxclient-devel mailing list
> Iax...@li...
> https://lists.sourceforge.net/lists/listinfo/iaxclient-devel
>   

[Iaxclient-devel] Iaxclient Browser Plugin

[Cristian S. <sep...@gm...>]

Hi Everyone,

We are developing RuMPI, a plugin for Browser to enable realtime
interaction, starting for VoIP using IAXClient, and later it will have
video&P2P.

This is not like SoftPhone, because all parameter are passed from server, so
you could control it from the app. (where to register and where to call), we
think  is perfect to enable click to call services using Asterisk.

If you are interested to develop or test it, feel free to contact me.

http://rumpi.sourceforge.net/

PS: Next week I'll go to web2expo.com and I'll stay for a Month in SFO,
could be really nice to meet face-to-face one of you and talk about
IaxClient, Asterix and other stuff.
-- 
Cristian Sepúlveda.

Te invito a visitar y comentar mis blogs:
http://www.osaka.cl
http://www.crsepulv.com

(56 9) 9 345 63 64.
MSN: crs...@ho...
LinkedIn.com: http://www.linkedin.com/in/cristiansepulveda

Re: [Iaxclient-devel] Problems with IAX-to-IAX calls

[Peter G. <jpg...@gm...>]

On Mon, Apr 14, 2008 at 11:17 AM, Ruslan Barski <ba...@gm...> wrote:
>  > I'm not sure I entirely understand the scenario. It sounds like you
>  > are saying there are clients A and B, where A is calling B. When A
>  > calls B, B sees an active|ringing state and then immediately
>  > transitions to just active. So far so good?
>
>  Correct!
>
>  > It's not clear to me how the ringing state would be changed in this
>  > case. In the iaxclient code, the ringing state only goes away when (1)
>  > the other end answers, (2) the other end is busy, (3) the other end
>  > sends a voice frame, or this end explicitly answers (calls
>  > iaxc_answer_call()).
>
>  No. Actually directly after we received the first correct state, there
>  is a change and the call state changes to state: active and
>  removed:ringing. But only when the caller is an IAX Client.

How do you know the other end did not send a voice frame?

>  First state: Correct
>  Added:  active ringing  OldState:       LocalCall: 0    State:  active ringing  Removed:
>
>  changes directly into
>  Second state:(Wrong?)
>  Added:  OldState:  active ringing       LocalCall: 0    State:  active  Removed:  ringing

Are you using one of the simpleclients such as testcall? What version
of iaxclient are you using?

>  > If the other end (client A) were to reject the call, the state should
>  > transition to free and no longer be either active or ringing. This is
>  > not what is happening?
>
>  No. See above!

A packet capture would be helpful. Do you know how to use tcpdump or wireshark?

Pete

Re: [Iaxclient-devel] iaxc_play_sound

[stevebvt <ste...@gm...>]

thanks for the info
i clearly missed the boat re: mix vs replace
& also that the behavior is not dependent on buffer full

i find it hard to believe that people don't miss a more
useful & reliable PlaySound mechanism

is there any sentiment to provide one?

steve
----- Original Message ----- 
From: "Peter Grayson" <jpg...@gm...>
To: "stevebvt" <ste...@gm...>
Cc: <iax...@li...>
Sent: Monday, April 14, 2008 10:42 AM
Subject: Re: [Iaxclient-devel] iaxc_play_sound


> On Mon, Apr 14, 2008 at 10:16 AM, stevebvt <ste...@gm...> wrote:
>> it's not a case of buffer overrun
>>
>>  what i was talking about was in the case of OutputBuffer full
>>  it seemed that (since there was only 1 OutputBuffer) that sound N+1 
>> would
>> overwrite
>>  sound N until the last sound was reached (which i considered 
>> problematic)
>
> In the case of multiple sounds, the several sounds are not
> overwritten, per se, they are mixed together.
>
>>  please explain the usefulness of the current functionality
>>  (clearly i'm missing it)
>
> I cannot justify the current functionality. I do not know why it mixes
> multiple sounds instead of playing them sequentially, but to keep the
> API stable, we cannot change the semantic of iaxc_play_sound().
>
>>  if i want to play N sounds (for example dtmf tones) & there is no way to
>> tell when a sound ends
>>  how do i send N distinct tones?
>
> For DTMF tones specifically, you can use iaxc_send_dtmf(). Multiple
> calls to that function will cause a sequence of IAX2 DTMF messages to
> be sent.
>
> Pete 

Re: [Iaxclient-devel] remote data is wrong

[Dennis C. <DCh...@pi...>]

Hi Peter,

On 14-Apr-08, at 10:55 AM, Peter Grayson wrote:

> Hi Dennis,
>
> On Mon, Apr 14, 2008 at 10:16 AM, Dennis Christopher
> <DCh...@pi...> wrote:
>> Hello,
>>
>>  I am attempting to implement local recording of the incoming audio
>>  stream in my iax client program.
>>  I am testing initially by recording the echo test.
>>
>>  I first set IAXC_AUDIO_PREF_RECV_REMOTE_RAW in the prefs, then my
>>  callback routine
>>  receives the struct iaxc_ev_audio audio, from which I reference the
>>  data pointer,
>>  and I write this out to a file. I'm expecting to receive 8khz 16-bit
>>  signed mono uncompressed data
>>  in LPCM format, as I gather from the list archives is correct.
>
> I think this is all right.
>
>>  The problem is that only the first 100 bytes or so contain data, the
>>  rest is null values.
>
> You will get one callback for each decoded audio packet. With speex,
> audio frames are sent every 20ms. That means 50 packets per second.
> 8kHz / 50 packets is 160 samples per packet. At two bytes per sample,
> you should be getting 320 bytes per callback. So indeed, 100 bytes
> does not seem right.
>
> There is a size field in the iaxc_ev_audio struct. What does it say
> the size of the data is?
>
> Pete

Thanks for the prompt reply.

It is as you calculated it.

The size field consistently informs me that I am receiving 320 bytes.

Dennis

Re: [Iaxclient-devel] remote data is wrong

[Peter G. <jpg...@gm...>]

Hi Dennis,

On Mon, Apr 14, 2008 at 10:16 AM, Dennis Christopher
<DCh...@pi...> wrote:
> Hello,
>
>  I am attempting to implement local recording of the incoming audio
>  stream in my iax client program.
>  I am testing initially by recording the echo test.
>
>  I first set IAXC_AUDIO_PREF_RECV_REMOTE_RAW in the prefs, then my
>  callback routine
>  receives the struct iaxc_ev_audio audio, from which I reference the
>  data pointer,
>  and I write this out to a file. I'm expecting to receive 8khz 16-bit
>  signed mono uncompressed data
>  in LPCM format, as I gather from the list archives is correct.

I think this is all right.

>  The problem is that only the first 100 bytes or so contain data, the
>  rest is null values.

You will get one callback for each decoded audio packet. With speex,
audio frames are sent every 20ms. That means 50 packets per second.
8kHz / 50 packets is 160 samples per packet. At two bytes per sample,
you should be getting 320 bytes per callback. So indeed, 100 bytes
does not seem right.

There is a size field in the iaxc_ev_audio struct. What does it say
the size of the data is?

Pete

Re: [Iaxclient-devel] iaxc_play_sound

[Peter G. <jpg...@gm...>]

On Mon, Apr 14, 2008 at 10:16 AM, stevebvt <ste...@gm...> wrote:
> it's not a case of buffer overrun
>
>  what i was talking about was in the case of OutputBuffer full
>  it seemed that (since there was only 1 OutputBuffer) that sound N+1 would
> overwrite
>  sound N until the last sound was reached (which i considered problematic)

In the case of multiple sounds, the several sounds are not
overwritten, per se, they are mixed together.

>  please explain the usefulness of the current functionality
>  (clearly i'm missing it)

I cannot justify the current functionality. I do not know why it mixes
multiple sounds instead of playing them sequentially, but to keep the
API stable, we cannot change the semantic of iaxc_play_sound().

>  if i want to play N sounds (for example dtmf tones) & there is no way to
> tell when a sound ends
>  how do i send N distinct tones?

For DTMF tones specifically, you can use iaxc_send_dtmf(). Multiple
calls to that function will cause a sequence of IAX2 DTMF messages to
be sent.

Pete

Re: [Iaxclient-devel] iaxc_play_sound

[stevebvt <ste...@gm...>]

it's not a case of buffer overrun

what i was talking about was in the case of OutputBuffer full
it seemed that (since there was only 1 OutputBuffer) that sound N+1 would 
overwrite
sound N until the last sound was reached (which i considered problematic)

please explain the usefulness of the current functionality
(clearly i'm missing it)

if i want to play N sounds (for example dtmf tones) & there is no way to 
tell when a sound ends
how do i send N distinct tones?

steve


----- Original Message ----- 
From: "Peter Grayson" <jpg...@gm...>
To: "stevebvt" <ste...@gm...>
Cc: <iax...@li...>
Sent: Monday, April 14, 2008 10:01 AM
Subject: Re: [Iaxclient-devel] iaxc_play_sound


> Hi Steve,
>
> On Thu, Apr 10, 2008 at 11:24 AM, stevebvt <ste...@gm...> wrote:
>>
>>
>> i have a question regarding multiple calls to iaxc_play_sound
>> (without waiting for a prior call to complete
>>  by the way - is there a way to determine when a call to iaxc_play_sound 
>> has
>> completed?)
>
> There is nothing in the API to know when a sound has completed. The
> api would have to be extended somehow to have this functionality. As
> it stands, iaxc_play_sound() has a fire-and-forget semantic.
>
>> i would have expected it to work this way
>> (given the case of a call to iaxc_play_sound - where a prior call has not
>> completed)
>>
>> 1. the new iaxc_sound{} would be added to the end of the existing chain 
>> of
>> iaxc_sounds (fifo)
>>    using iaxc_sound.next
>> 2.when the current iaxc_sound completed the next iaxc_sound would become
>>    the head of the fifo queue & output
>>
>> what the code sees to do is
>> 1. the new iaxc_sound{} is added to the front of the existing chain of
>> iaxc_sounds (lifo)
>>     pa_play_sound () -
>>       sound->next = sounds;
>>       sounds = sound;
>> 2. if there is BUFFER OVERFLOW subsequent iaxc_sound{}  data seems to be
>> discarded
>>    pa_mix_sounds () -
>>      if ( *sp ) /* don't advance if we removed this member */
>>       sp = &((*sp)->next);
>>
>> what should happen in this case?
>
> The way iaxc_play_sound() works is to play each sound as soon as
> possible. If there are multiple sounds in the "sounds" list, then they
> are played at the same time and mixed together (see mix_slin()). This
> is why it doesn't matter whether the sounds are prepended or appended
> to the sounds list.
>
> I do not see any potential buffer overflows in pa_mix_sounds(), but I
> could be missing what you are seeing. Could you give a little more
> explanation of where a buffer may be overrun?
>
> Pete 

[Iaxclient-devel] remote data is wrong

[Dennis C. <DCh...@pi...>]

Hello,

I am attempting to implement local recording of the incoming audio  
stream in my iax client program.
I am testing initially by recording the echo test.

I first set IAXC_AUDIO_PREF_RECV_REMOTE_RAW in the prefs, then my  
callback routine
receives the struct iaxc_ev_audio audio, from which I reference the  
data pointer,
and I write this out to a file. I'm expecting to receive 8khz 16-bit  
signed mono uncompressed data
in LPCM format, as I gather from the list archives is correct.

The problem is that only the first 100 bytes or so contain data, the  
rest is null values.

Dennis Christopher
Pixion Inc.

Re: [Iaxclient-devel] iaxc_play_sound

[Peter G. <jpg...@gm...>]

Hi Steve,

On Thu, Apr 10, 2008 at 11:24 AM, stevebvt <ste...@gm...> wrote:
>
>
> i have a question regarding multiple calls to iaxc_play_sound
> (without waiting for a prior call to complete
>  by the way - is there a way to determine when a call to iaxc_play_sound has
> completed?)

There is nothing in the API to know when a sound has completed. The
api would have to be extended somehow to have this functionality. As
it stands, iaxc_play_sound() has a fire-and-forget semantic.

> i would have expected it to work this way
> (given the case of a call to iaxc_play_sound - where a prior call has not
> completed)
>
> 1. the new iaxc_sound{} would be added to the end of the existing chain of
> iaxc_sounds (fifo)
>    using iaxc_sound.next
> 2.when the current iaxc_sound completed the next iaxc_sound would become
>    the head of the fifo queue & output
>
> what the code sees to do is
> 1. the new iaxc_sound{} is added to the front of the existing chain of
> iaxc_sounds (lifo)
>     pa_play_sound () -
>       sound->next = sounds;
>       sounds = sound;
> 2. if there is BUFFER OVERFLOW subsequent iaxc_sound{}  data seems to be
> discarded
>    pa_mix_sounds () -
>      if ( *sp ) /* don't advance if we removed this member */
>       sp = &((*sp)->next);
>
> what should happen in this case?

The way iaxc_play_sound() works is to play each sound as soon as
possible. If there are multiple sounds in the "sounds" list, then they
are played at the same time and mixed together (see mix_slin()). This
is why it doesn't matter whether the sounds are prepended or appended
to the sounds list.

I do not see any potential buffer overflows in pa_mix_sounds(), but I
could be missing what you are seeing. Could you give a little more
explanation of where a buffer may be overrun?

Pete

[Iaxclient-devel] Problems with IAX-to-IAX calls

[Ruslan B. <ba...@gm...>]

Hello,

when we call from IAX-client to a IAX-client, we get a problem in the
call-life-cycle.
Normally when a new call comes in, it has the state "active ringing".
In our case the state is changing. We get 2 states in sequence, first
state is ok (active ringing), but then the call changes to state is
active, removed is ringing.
Therefor we can't discover if the call was rejected by remote before
we accepted the call.
Thus our client shows an active call when the call is allready canceled.

Any idea?

[Iaxclient-devel] Problem with iaxc_register

[Jean-François G. <jf....@sq...>]

Hi,

I don't understand how to register to an iax server an place a call.
For exemple trying with testcall program gived with sources.

Supposing i've a peer in my iax.conf :

[jf]
type=friend
secret=foo
auth=plaintext
host=dynamic
context=internal
peercontext=internal
qualify=yes


and in my extensions.conf

[internal]
exten => 888,1,Dial(IAX2/888@10.0.0.4)

When trying with testcall :

./testcall -u jf -p foo -h 10.0.0.2 888

OR :
./testcall -u jf -p foo -h 10.0.0.2 888@internal

i see in my cli :
: chan_iax2.c:6027 update_registry: Restricting registration for peer 
'jf' to 60 seconds (requested 300)

But no call.. no sound.. nothing happen.

If i try :
./testcall -u jf -p foo -h 10.0.0.2 888@10.0.0.4

It runs very well...

I don't understand why i can't use exten of my context and how to avoid 
this strange behavior.

regards,

jfg