space88man left a comment (kamailio/kamailio#4800)
Let me at least address this:
> This PR is monstrous in scope, and introduces lots of new constructs.
This is not the case at all:
- it heavily reuses everything that kamailio does well - the io_wait loop and
the SIP parser
- it fact in the TCP worker it simplifies it: the workers no longer have to
deal with - am I receiving a file descriptor or am I receiving network data.
The worker only receives a completely prepared SIP message task (the actual SIP
message lies in shared memory).
- instead of chunking SIP data in the worker we do that in PROC_TCP_MAIN using
exactly the same code from `tcp_read.c`
- note that the write path always had 1. worker writes or 2. PROC_TCP_MAIN
writes - this was awkward. Now there is only one write path - PROC_TCP_MAIN is
doing the writes.
- there was a lot of complexity around sendmsg and fd-passing. All this is
gone in reactor mode.
--
Reply to this email directly or view it on GitHub:
https://github.com/kamailio/kamailio/pull/4800#issuecomment-4856511969
You are receiving this because you are subscribed to this thread.
Message ID: <kamailio/kamailio/pull/4800/[email protected]>
_______________________________________________
Kamailio - Development Mailing List -- [email protected]
To unsubscribe send an email to [email protected]
Important: keep the mailing list in the recipients, do not reply only to the
sender!
