[asterisk-dev] t.38 passthrouth

[Steve Underwood]

Johansson Olle E wrote:

>
> 13 mar 2006 kl. 16.27 skrev anthony thomas:
>
>> Hello all,
>>  We are testing the t38passthrough branch in our
>> gateways and can not fully understand why the RTP
>> stream has to go througth the * box disabling native
>> RTP bridging.
>>
>>  What is the problem in allowing a second RE-INVITE to
>> switch to T.38?
>
>
> I asked Steve the same question and got the answer that
> "It was mostly done for testing..." so I guess if you want to
> disable the code and test without the sip_bridge(), you can
> do that. Please confirm your findings to the bug tracker, so
> we can make a decision whether this is the way forward
> - to always run the rtp bridge and thus allow re-invites.

This is correct. We fudged this to force the data through the * box, so 
we could test the handling of passthrough. Most things were simply 
reinviting and bypassing the box. It should be changed now, to allow 
direct peer-to-peer communication. It might, however, be worth keeping 
it as a compile time option to simplify further testing of passthrough.

Regards,
Steve