> On 12/01/2014 20:51, Theo de Raadt wrote: > > And now let's ask the question which matters. What usage patterns do > > you break with this change? > > > > Did you think of that? Or are you only thinking of your own uses..? > > Do you have any particular and practically important failing scenario in > mind?
Millions of users other than you. > The change in only for AF_INET, and only the move of > close(listen_socket) up, which was closed after work, and now is closed > before work. Only for AF_INET, you mean, the most common case? > The failing case would be when second user would be trying to connect > while the first user is communicating, and would now get immediate > failure to connect, vs. before would have gotten > connect+wait+disconnect. I don't see this how this difference can be > practically important. Wow..
