On Thu, 26 Nov 2020 12:38:24 +0100 Christian Grothoff <groth...@gnunet.org> wrote:
> On 11/26/20 8:31 AM, José Bollo wrote: > > On Tue, 24 Nov 2020 23:12:43 +0100 > > Christian Grothoff <groth...@gnunet.org> wrote: > > > >> On 11/24/20 4:50 PM, José Bollo wrote: > >>> What about if in some future an option to MHD_upgrade_action would > >>> enable to "extract" the socket? Something like: > >>> > >>> MHD_upgrade_action (urh, MHD_UPGRADE_ACTION_OWN_SOCKET); > >>> > >>> with a clear semantic on the implications that it has on > >>> half-closing. > >> > >> I'm not sure it's needed. What's wrong with dup()? > >> > > > > Something is wrong with dup: it needs an other file descriptor, > > dividing by 2 the count of available connection. > > For child processes, not really: you can even do the dup()ing only > after fork() if you manually clear the close-on-exec flag. So you > don't need any extra FDs if you do it 'just right'. Yeah but fork+exec is a requirement > Besides, with epoll() and a proper ulimit setup, you virtually have no > limit on # connections these days anyway. for sure unlimited resource on unlimimited earth -_-
