Test suites are more reliable (pesky EADDRINUSE bugs are hopefully squashed now) Thenng_closefunction is renamed tonng_sock_close, as a heads up... so pay attention to the migration guide! (There is a macro,NNG1_MIGRATIONthat can be set to provide a macro to aid in transition. Summary ...
123 119 NUTS_PASS(nng_aio_alloc(&aio1, NULL, NULL)); 124 120 NUTS_PASS(nng_aio_alloc(&aio2, NULL, NULL)); @@ -147,11 +143,11 @@ test_udp_multi_send_recv(void) 147 143 NUTS_PASS(nng_aio_set_iov(aio4, 1, &iov4)); 148 144 NUTS_PASS(nng_aio_set_input(aio4, 0...
gdamorecommentedNov 26, 2023 I believe that the problem here is that the protocols that use transportpipe_sendcan wind up losing the message if the pipe has been closed. This is because the transport uses nng_aio_begin (checking for success), but doesn't complete the aio (we need tonot...
if ((probe != NULL) && (item->aio != NULL)) { struct sockaddr_in *sin; struct sockaddr_in *sin; #ifdef NNG_ENABLE_IPV6 struct sockaddr_in6 *sin6; nng_sockaddr *sa = item->sa; #endif nng_sockaddr *sa = item->sa; switch (probe->ai_addr->sa_family) { case AF_INET: @@...
nni_sleep_aio(s->retry, &p->time_aio); return; }// start message resending // msg = nni_id_get_min(&p->sent_unack, &pid); // if (msg != NULL) { @@ -753,6 +790,9 @@ mqtt_recv_cb(void *arg) // schedule another receive nni_pipe_recv(p->pipe, &p->recv_aio);/...
fix aio test failures on darwin … Verified 86d9b19 View details gdamore merged commit 73dbf23 into stable Jan 3, 2025 16 checks passed gdamore deleted the stable-socket-close branch January 3, 2025 17:24 BrewTestBot mentioned this pull request Jan 5, 2025 nng 1.10 Homebrew/homebr...