On Jul 17, 2008, at 2:35 PM, Brian Aker wrote:
What I don't quite understand is, why use non-blocking I/O?
This would be like polling the connection. But after setting a
timeout you shouldn't need to poll.
Much cheaper though then the current method. Let me describe more
detail...
Another thing:
Have you considered using c10k and a pool of threads?
Notice that in Drizzle, we only work with pool of threads.
I have tree for this, but it is now out of date, but this was the
plan...
Pool of threads (done).
Change main select to be libevent (done in different tree).
Executue on read events. If we ever get a zero read on a non-block
call (everything is non-block) we yield after N events and wait till
we get sliced back in. This would be done on read/writes (though
just read first).
Thoughts?
OK, I see. Because libevent has already detected that data is
available, the thread does a "busy" read.
Is this faster than doing a blocking read?
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
