[Webkit-unassigned] [Bug 61287] [UNIX] Use SOCK_SEQPACKET when available

Sun May 29 23:10:25 PDT 2011

https://bugs.webkit.org/show_bug.cgi?id=61287

--- Comment #35 from Kimmo Kinnunen <kimmo.t.kinnunen at nokia.com>  2011-05-29 23:10:25 PST ---
(In reply to comment #34)
> My opinion (and not more) after reading the discussion. If we want to support systems without SOCK_SEQPACKET (and I guess we do) then we need to implement the fallback for SOCK_DGRAM.
> So, I don't think we should implement a path for SOCK_SEQPACKET at all (for systems where it is available).

I don't really understand these two comments.
There's no separate codepath for SOCK_SEQPACKET. The main codepath for SEQPACKET will be identical to DGRAM, eg. you can just switch the socket type and it will work. 

The difference with it is that with SEQPACKET you can observe the closing of the stream. And here is where you need the fallback if DGRAM is used. And the fallback can be an extra dummy stream socket.

> Furthermore, if SOCK_DGRAM is more effective than SOCK_SEQPACKET (I guess Kimmo and Simon believes that it is) than it is better to use SOCK_DGRAM with the fallback everywhere.

No, this is wrong interpretation. I never said that there would be a perf difference in SEQPACKET vs. DGRAM. In my comments "datagram" is used to define the type of the message handling, as in as opposed to "stream". IOW: SEQPACKET == datagram.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.