[asterisk-dev] proposed change to main/http.c

[Kevin P. Fleming]

----- Luigi Rizzo <rizzo at icir.org> wrote:
> It shouldn't harm on linux either, as the socket should really
> be blocking to make input processing correct and efficient

I don't understand why the main socket used here is non-blocking at all; it makes no sense. If it wasn't, then the sub-sockets created by accept() would be blocking by default, and everything would be happy.

I'll go ask Mark (somewhat politely) what he was thinking :-)

-- 
Kevin P. Fleming
Senior Software Engineer
Digium, Inc.