> Is the transport type used for one transaction > in a call to be the same in other transactions > of the same call?
It can be. However it is not required. > Alan Johnston's "SIP: Understanding the session initiation > protocol" further suggests that each TCP connection > should be successively closed/re-opened per transaction. I am not sure what is communicated within the book. However the current direction of the SIP community is to leave the connection open until it needs to be reclaimed. > Presuming that my understanding of the RFC3261 and A. > Johnston's book to be correct (transport type is a per > transaction property), is there any mechanism in SIP of > forcing an entire call to use the same transport type > for all it's transactions. Using SIPS can have the characteristic of forcing tls. Otherwise, I am not aware of any RFC or draft which requires the same transport end-to-end for all requests within the dialog. And devices should not indicate the "current" transport within a Contact or Record-Route entry unless it knows the transport is supported and/or is requiring it to be used. Note that proxies might not record-route and remain within the dialog. And the device on the other side of the proxy might not support (or be willing use) the indicated transport. _______________________________________________ Sip-implementors mailing list [email protected] http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
