On Tue, Feb 10, 2009 at 12:22 AM, D. S. <[email protected]> wrote:
> On Sun, Feb 8, 2009 at 9:28 PM, Alexander Chemeris
> <[email protected]> wrote:
>> On Fri, Feb 6, 2009 at 11:52 PM, D. S. <[email protected]> wrote:
>>> As it stands, G.722 sessions will not connect using Recon with sipXtapi.
>>> The reason is that G.722 sessions are negotiated with 8000 as the
>>> sample rate even though it actually operates at 16000.
>>> See sec. 4.5.2 in RFC3551 and the following link:
>>> http://en.wikipedia.org/wiki/G.722
>>
>> Yes, this is a known problem, and it's already workarounded in
>> sipXtapi. Look at following lines:
>> sipXtackLib\src\net\SdpBody.cpp:1272
>> sipXtackLib\src\net\SdpBody.cpp:1778
>> sipXtackLib\src\net\SdpHelper.cpp:220
>>
>> So if it does not work with recon, it's a problem with recon I guess.
>> I recall I tested this with some phones and it worked.
>
> Thank you for pointing me to this.
> This would require Recon to hack the sample rate passed to sipXsdpLib
> because Recon uses Resiprocate for the SIP stack.
> IMHO, it would be better to make changes to the static tables in
> SdpDefaultCodecFactory.cpp and plgg722.c because it would be handled
> by the SDP library itself for all calling SIP stacks.
> Static data changes are simpler than coded if-thens.
> Of course, your mileage may vary if something depends on this somehow
> and breaks.

Yeah, I recall it was a deliberate decision. I recall value from
the SDP is used to determine how to handle this codec in
flowgraph. E.g. if reported codec samplerate differs from
a flowgraph samplerate, data will be automatically
up/downscaled by decoder/encoder resource. (you need
to compile with Speex support enabled for this).

So, it's a bug in specs and it should be handled as a bug.
I.e. it should introduce as small confusion as possible.
My hacks are small enough, and I hope recon can be
hacked the same way.

-
Regards,
Alexander Chemeris.

SIPez LLC.
SIP VoIP, IM and Presence Consulting
http://www.SIPez.com
tel: +1 (617) 273-4000
_______________________________________________
sipxtapi-dev mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipxtapi-dev/

Reply via email to