Kamil Dudka wrote: > I'd abstain from driving the development by observation of existing > systems. > The specification compliance is the way to go.
Yes, I agree. > > To me, the behavior you say is natural seems to be in direct > > conflict with the documentation. > > What exactly is in the conflict? I have always understood select(2) (and later select_tut(2)) such that if select() returns without error and FD_ISSET(socket,set) then the corresponding io operation on socket can reliably be assumed to not block, also when socket is in blocking mode. Daniel says that the io operation can in fact block anyway, when the socket is in blocking mode. To me, this would seem to make select() useless for blocking io, which I find quite surprising. //Peter _______________________________________________ libssh2-devel http://cool.haxx.se/cgi-bin/mailman/listinfo/libssh2-devel
