Re: [reSIProcate] Dynamic binding to interfaces and ports
- From: Alan Hawrylyshen <alan@xxxxxxxxxx>
- Date: Thu, 27 Oct 2005 12:57:25 -0600
On 27-Oct-05, at 12:01 , Matthias Moetje - TERASENS GmbH wrote:
Jason,
thanks very much for your reply.
yes. this is a good idea. it is non-trivial to implement since
SipMessages can have pointers to Transports. We would need to switch
to using TransportHandles to solve this issue. Additionally, the
removeTransport will need to be made thread-safe. Ideally, removing a
transport will happen when that Transport is no longer being used by
any active transactions.
OK. What about marking a transport as disabled instead? Would this
be easier? The transport object could continue to exist but the
TCP/IP binding would be removed...?
Sounds like a leak recipe. I think we can get away with reference
counting the transports.
When it drops to zero, and a removal is requested, it is removed.
Thoughts?
A