On Wed, Jun 20, 2012 at 7:52 AM, Lauri Kasanen <[email protected]> wrote: > On Wed, 20 Jun 2012 07:46:05 -0600 > Eduardo Silva <[email protected]> wrote: > >> > I'm not 100% sure where the event_close comes from (maybe client pressed >> > stop, etc etc), but the first event_close is valid. The false event_write >> > after that is not. >> >> but be aware of the following, if your plugin request to close the >> connection on that same epoll loop a write event could be already >> en-queued, i am trying to reproduce the issue in the CGI context. > > It's possible, yeah. > > But given that the connection is closed, the cgi plugin no longer has the > data on which fd it was, and so can't tell the write event was meant for that. > > In that context can't think of a better solution; I also don't see any good > from trying to close a connection that doesn't have a cs, wouldn't a missing > cs always mean the connection was already closed? > > - Lauri
I have reviewed the issue and the solution is correct, no other information exists about the socket in question and returning 0 will not hurt anyone, it "seems" like the hooks are being called in the previous close hooks so its ok. applying patch, thanks! -- Eduardo Silva http://edsiper.linuxchile.cl http://www.monkey-project.com _______________________________________________ Monkey mailing list [email protected] http://lists.monkey-project.com/listinfo/monkey
