2011/5/6 Raphael Coeffic <[email protected]>:
> On 06.05.11 23:14, Iñaki Baz Castillo wrote:
>>
>> This is, such ACK is matched
>> by the transport layer so it should never arrive to core/UAS layer in
>> which dialogs exist.

Sorry, I meant "the transaction layer".


> In fact, ACKs are matched by the transaction layer in SEMS. Last time I
> looked at the code, I was also wondering whether or not non-2xx-ACK should
> really be passed to the UA layer.

They should not :)


> Would anybody oppose a patch which would aborb non-200-ACKs, instead of
> passing them to the UA layer?

Not me.


> Or asked differently: can anybody think of a reason for passing those
> non-200-ACKs to the application?

When a UAS or a proxy replies a [3456]XX response for an INVITE, the
core/UA/application logic is done. The arriving ACK never contains
information useful for application layer as it's hop by hop (so it
doesn't come frrom the call originator) and it's just a message to
complete de 3-steps-handshake at transport level (in case of an ACK or
2XX, it completes the handshake at application/core/UA level which is
very different).

 So, for sure an ACK for [3456]XX should never arrive to the core
layer. RFC 3261 also states it.

-- 
Iñaki Baz Castillo
<[email protected]>
_______________________________________________
Sems mailing list
[email protected]
http://lists.iptel.org/mailman/listinfo/sems

Reply via email to