> I was looking at the conversation code to use it in a rlogin
> dissector I'm working on. I see that the conversation_t type
> data pointer isn't destroyed in the conversation_init routine.
I assume you're referring here to the "data" member of a
"conversation_t".
> Is there a reason for this?
The conversation code has no clue where the data pointer points - it
might not point to allocated memory at all, or it might point to memory
allocated with "malloc()"/"g_malloc()", or it might point to memory
allocated with "g_mem_chunk_alloc()", or....
As such, it has no clue how to free it.
> Whichever way is picked; it needs to be documented in the
> developers guide.
Yes, the developer's guide section on the conversation code could should
have that.
Of course, it'd be useful if the developer's guide section on the
conversation code had something that mentioned that you can attach a
data pointer to the conversation data structure in the first place -
it'd be useful, in fact, if it had more than
2.2 Following "conversations."
In ethereal a conversation is ...
2.2.1 The conversation_init function.
2.2.2 The conversation_new function.
2.2.3 The find_conversation function.
2.3 ??
but that's all it has now, so it's not as if there's enough there to
provide a place to which to add a discussion of the data pointer....
