Just wondering: would it make sense to provide a libevent API for this? I ask because proper shutdown is important to a number of us, and it seems a shame that we all have to independently sprinkle the code throughout our programs wherever sockets are used. On the surface, at least, it would appear that a simple API could take care of this for us.
On Oct 30, 2010, at 9:55 AM, Kelly Brock wrote: > Hi Nick, > >>> Ok, so figured out a way to do this and it seems clean and works with >> the >>> select based backend but not the iocp based backend. Basically I just >> set >>> >>> But, it does not seem to work on the IOCP version which seems a bit odd >>> since this should be handled at the OS level and automatically dealt >> with. >>> I'm fairly confident of the test code given it starts and stops >> event_base >>> 10 times, runs 1024 sockets through the connection, send/recv and >> shutdown >>> all without errors. >>> >>> I'll see if I can figure out the problem but it's been a while since I >> poked >>> the hornet's net which is IOCP... >> >> Hi, Kelly, and thanks for checking that out. > > Just a note before the details. I would suggest adding a little > section to the manual covering the proper method of handshaked shutdown. > It's just a minor thing but figuring out the watermark method is non-obvious > till you give it some further thought. (Well, it was to me at the time, > then again it was 2am. :)) > > Anywhere, here is a quick and dirty testbed for the problem. > > http://pastebin.com/QF4N2DEw > > If you leave "USE_IOCP" set to 0 everything shuts down correctly. > If you turn that flag on though, the EOF's are never seen. I just got this > test case functional so forgive the dirty leaky implementation if you will, > but please do check that I'm not just making some other stupid error as I'm > still getting used to 2.0 changes. I'm sure there are other problems beyond > the test case in this code; I put it together relatively quickly by > extracting bits and pieces out of my C++ wrappers/abstraction. > >> If you don't find the bug (or if you don't find a fix for the bug) >> right away, could you please post some test code to help reproduce and >> demonstrate it, either here or on the bugtracker? (Send a link if >> it's long.) This is the kind of thing we should get fixed in 2.0 if >> possible. > > I just got this functional so haven't started debugging. If I > happen to find it in the next 30 minutes before I run out will post a fix, > if not I will be unlikely to get back to this till later tonight. > > KB > > > *********************************************************************** > To unsubscribe, send an e-mail to majord...@freehaven.net with > unsubscribe libevent-users in the body. *********************************************************************** To unsubscribe, send an e-mail to majord...@freehaven.net with unsubscribe libevent-users in the body.
