[Guy Harris <gharris@xxxxxxxxx>]

Peter Johansson wrote:

> Yes, in this case all future connections made to 192.168.169.2/2441 
> should be handled by the same protocol dissector.
> Do you (or anyone else for that matter) have an idea of how to 
> accomplish the "template" conversation you had in mind?
> I got the idea that one could add another option value which could be 
> used in conjunction with either NO_ADDR_B, NO_PORT_B or both, say 
> PERSISTENT (value should be 0x08 or perhaps 0xF0).

I'm not sure "persistent" is the right adjective, but I'm not sure I 
have any better ideas....  "Prototype"?  "Template"?  The idea is that 
it's not creating a regular conversation, it's creating a prototype or 
template from which one or more templates

> The dissector could 
> then do:
> c = conversation_new(addr1, addr1, PTYPE, port1, 0, NO_ADDR_B | 
> NO_PORT_B | PERSISTENT).
> When find_conversation completes the conversation for a newly made 
> connection for c, it should instead of g_hash_table_remove do 
> g_hash_table_insert for the established connection when options specify 
> PERSISTENT in conjunction with either of (or both) NO_ADDR_B and 
> NO_PORT_B..

I.e., when it completes the conversation, it should add a new 
non-persistent, non-wildcard conversation?

That sounds like the right thing to do.