On Fri, 17 Mar 2000, Alex Korobka wrote:

>   OK. Makes sense. Still, I'd rather avoid checking all sockets explicitly 
>   in the sock_reselect(). This will result in duplicate notifications for
>   sockets which are still in the server poll fd set.

No, the sock_poll_event() should clear the poll mask for any notification
it issues (the hmask masks them out), so the second call (which would use
the new mask) won't reissue them. (The recursive sock_reselect should be
harmless because of this)

Reply via email to