----Priorité élevée------------------------ - CGI (TODO HUGO) - Need to test normal body parsing (Verif avec HUGO) - curl --resolve, for testing hostname curl -v --resolve server1:4040:127.0.0.1 --resolve server2:4040:127.0.0.1 server1:4040 - test limit de connexions sur listen() - handle redirection (Work, but weird behavior need deeper test) - Ecrire des tests ! - cgi_cpp_status.cpp with POST dont show error page. Normal or not ? For non blocking CGI : // We could maybe, // add FD_RD_FR_CHLD to epoll, // return to the main loop, // read FD_RD_FR_CHLD each time epoll say its ready, // then try waitpid() with WNOHANG after each read. // when waitpid() tell us its finish (or maybe when epoll return EPOLLHUP) // then actually parse the script_output and send it to the client. - check status in autoindex - merge changes from hugo5 to master (attention a pas casse svp :clown:) ----Priorité modérée------------------------ - namespace utils ? - change "std::string" to reference "std::string &" in most functions and add "const" if apropriate. - peut-être check si ip > 32bits ----Priorité faible------------------------ - idealy, we should not clear() raw_request after a response, but just the part we actually parsed for this response. I think the client could send multiples request on the same connection one after the other without waiting for response. So raw_request could contain more than the first request we handle. - chunked request (need testing) - client_body_limit 0 valeur special pour desactiver dans config - gerer le champ "Accept" du client - gerer les ".." dans un URL (verifier que l'on ne sort pas du dossier "root") - do correct handling of special character in url (/rfc2119_files/errata.js.t%C3%A9l%C3%A9chargement -> /rfc2119_files/errata.js.téléchargement) - maybe add a "last_action_time" in Client for timeout handling little global timeout on epoll, like 100ms, then find client that actualy need to timeout if (actual_time - client.last_action_time > 10000ms){timeout(client)} - add headers "Date" and "Last-Modified" to response - change "std::string" to reference "std::string &" in most functions and add "const" if apropriate. - Il faut vérifier le path de la requête, voir si le serveur est bien censé délivrer cette ressource et si le client y a accès, avant d'appeler le CGI. Valgrind error RESOLVED ! : ==847174== 1,314 bytes in 1 blocks are definitely lost in loss record 1 of 1 ==847174== at 0x483BE63: operator new(unsigned long) (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so) ==847174== by 0x49AA35D: std::__cxx11::basic_string, std::allocator >::_M_mutate(unsigned long, unsigned long, char const*, unsigned long) (in /usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.28) ==847174== by 0x49ABB52: std::__cxx11::basic_string, std::allocator >::_M_append(char const*, unsigned long) (in /usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.28) ==847174== by 0x426BC5: Webserv::_read_cgi_output(cgi_pipe_rfd&) (run_loop.cpp:39) // (LUKE: its an string.append() call) ==847174== by 0x427962: Webserv::run() (run_loop.cpp:153) ==847174== by 0x4052E9: main (main.cpp:38) RESOLVED, AGAIN :D ! --loop epoll() EPOLLHUP on client fd 8 47067----loop epoll() EPOLLHUP on client fd 8 47068----loop epoll() EPOLLHUP on client fd 8 47069----loop epoll() EPOLLHUP on client fd 8 47070----loop epoll() EPOLLHUP on client fd 8 47071----loop epoll() EPOLLHUP on client fd 8 47072----loop epoll() EPOLLHUP on client fd 8 47073----loop epoll()