Re: bug#62044: 30.0.50; ERC 5.5: Auto-reconnect is broken
"J.P." <[email protected]>
| Newsgroups | gmane.emacs.erc.general |
|---|---|
| Message-ID | <[email protected]> |
"J.P." <[email protected]> writes: > I've attached a less sloppy version that probably still fails in some > common cases, but at least it reuses the existing session connector. I've improved upon this further (v3 attached) by adding a housekeeping task to monitor the initial server process from creation. Such a move may be regrettable because it adds yet more complexity to the already dizzying auto-reconnect landscape. However, I couldn't find a suitable way to cover common process errors that aren't presented to the sentinel but still need to engage the reconnect logic. If this leads to a futile game of whack-a-mole, we'll obviously need to try a different approach. But if we do more-or-less build on what I've got so far, we'll definitely need to ensure it agrees with 27 and 28 before spending serious energy on refinement and tests. Thanks.
0000-v2-v3.diff
(text/x-patch, 6.1 KB)
From 7dcccaef52ad4442eabaabe04cd95ffceec048f7 Mon Sep 17 00:00:00 2001 From: "F. Jason Park" <[email protected]> Date: Thu, 9 Mar 2023 06:33:38 -0800 Subject: [PATCH 0/1] *** NOT A PATCH *** *** BLURB HERE *** F. Jason Park (1): Add conditional erc-server-reconnect-function lisp/erc/erc-backend.el | 93 ++++++++++++++++++++++++++++++++++++++++- 1 file changed, 91 insertions(+), 2 deletions(-) Interdiff: diff --git a/lisp/erc/erc-backend.el b/lisp/erc/erc-backend.el index d289df98bab..1030672506e 100644 --- a/lisp/erc/erc-backend.el +++ b/lisp/erc/erc-backend.el @@ -658,6 +658,31 @@ erc--register-connection (run-hooks 'erc--server-post-connect-hook) (erc-login)) +(defvar erc--server-connect-function #'erc--server-propagate-failed-connection + "Function called one second after creating a server process. +Called with the newly created process just before the opening IRC +protocol exchange.") + +(defun erc--server-propagate-failed-connection (process) + "Ensure the PROCESS sentinel runs at least once on early failure. +Act as a watchdog timer to force `erc-process-sentinel' and its +finalizers, like `erc-disconnected-hook', to run when PROCESS has +a status of `failed' after one second. Print the +`process-exit-status', which can be a number, like 111, or a +message, like \"TLS negotiation failed\"." + (when (eq (process-status process) 'failed) + (erc-display-message + nil 'error (process-buffer process) + (format "Process exit status: %S" (process-exit-status process))) + (pcase (process-exit-status process) + ((guard erc--server-reconnect-timer)) + (111 + (delete-process process)) + (`(file-error ,(rx "failed") ,(rx "unreachable") . ,_) + (erc-process-sentinel process "failed with code -523\n")) + ((rx "tls" (+ nonl) "failed") + (erc-process-sentinel process "failed with code -525\n"))))) + (defvar erc--server-connect-dumb-ipv6-regexp ;; Not for validation (gives false positives). (rx bot "[" (group (+ (any xdigit digit ":.")) (? "%" (+ alnum))) "]" eot)) @@ -708,9 +733,11 @@ erc-server-connect (with-current-buffer buffer (erc-current-nick)))) ;; wait with script loading until we receive a confirmation (first ;; MOTD line) - (if (eq (process-status process) 'connect) + (if (process-contact process :nowait) ;; waiting for a non-blocking connect - keep the user informed - (erc-display-message nil nil buffer "Opening connection..\n") + (progn + (erc-display-message nil nil buffer "Opening connection..\n") + (run-at-time 1 nil erc--server-connect-function process)) (message "%s...done" msg) (erc--register-connection)))) @@ -759,9 +786,9 @@ erc-server-delayed-check-reconnect 2))) (let ((reschedule (lambda (proc) - (let ((erc-server-reconnect-timeout - erc--server-reconnect-timeout)) - (with-current-buffer buffer + (with-current-buffer buffer + (let ((erc-server-reconnect-timeout + erc--server-reconnect-timeout)) (delete-process proc) (erc-display-message nil 'error buffer "Nobody home...") (erc-schedule-reconnect buffer 0)))))) @@ -770,18 +797,22 @@ erc-server-delayed-check-reconnect "*erc-connectivity-check" nil erc-session-server erc-session-port :nowait t)) - tls-check) - (when (and (not (eq erc-session-connector - #'erc-open-network-stream)) - (process-contact proc :tls-parameters)) - (setq tls-check - (run-at-time - 1 1 (lambda (proc) - (unless (eq 'connect (process-status proc)) - (cancel-timer tls-check)) - (when (eq 'failed (process-status proc)) - (funcall reschedule proc))) - proc))) + (check-expire (time-add 10 (current-time))) + check-aside) + (setq check-aside + (run-at-time + 1 1 (lambda (proc) + (when (or (not (eq 'connect (process-status proc))) + (time-less-p check-expire (current-time))) + (cancel-timer check-aside)) + (when (time-less-p check-expire (current-time)) + (erc-display-message + nil 'error buffer "Timed out waiting on `connect'") + (delete-process proc) + (funcall reschedule proc)) + (when (eq 'failed (process-status proc)) + (funcall reschedule proc))) + proc)) (set-process-filter proc (lambda (proc _) (delete-process proc) @@ -790,14 +821,16 @@ erc-server-delayed-check-reconnect (run-at-time nil nil #'erc-server-delayed-reconnect buffer))) (set-process-sentinel - proc (lambda (cproc event) + proc (lambda (proc event) (with-current-buffer buffer (pcase event ("open\n" - (run-at-time nil nil #'send-string - cproc "PING *connect-check*\r\n")) - ("connection broken by remote peer\n" - (funcall reschedule cproc))))))) + (run-at-time + nil nil #'send-string proc + (format "PING %d\r\n" (time-convert nil 'integer)))) + ((or "connection broken by remote peer\n" + (rx bot "failed")) + (funcall reschedule proc))))))) (file-error (funcall reschedule nil)))))) (defun erc-server-filter-function (process string) -- 2.39.2
0001-Add-conditional-erc-server-reconnect-function.patch
(text/x-patch, 6.4 KB)
From 7dcccaef52ad4442eabaabe04cd95ffceec048f7 Mon Sep 17 00:00:00 2001 From: "F. Jason Park" <[email protected]> Date: Wed, 8 Mar 2023 06:14:36 -0800 Subject: [PATCH 1/1] Add conditional erc-server-reconnect-function * lisp/erc/erc-backend.el (erc--server-reconnect-timer, erc-server-delayed-check-reconnect): Add possible alternate value for option `erc-server-reconnect-function' that only attempts to reconnect after hearing back from the server. Also add helper variable. (erc-server-connect): Run erc--server-connect-function for async processes. (erc--server-connect-function, erc--server-propagate-failed-connection): Add connect-function variable and default value to run on process creation and execute process-failure handlers. --- lisp/erc/erc-backend.el | 93 ++++++++++++++++++++++++++++++++++++++++- 1 file changed, 91 insertions(+), 2 deletions(-) diff --git a/lisp/erc/erc-backend.el b/lisp/erc/erc-backend.el index 567443f5329..1030672506e 100644 --- a/lisp/erc/erc-backend.el +++ b/lisp/erc/erc-backend.el @@ -658,6 +658,31 @@ erc--register-connection (run-hooks 'erc--server-post-connect-hook) (erc-login)) +(defvar erc--server-connect-function #'erc--server-propagate-failed-connection + "Function called one second after creating a server process. +Called with the newly created process just before the opening IRC +protocol exchange.") + +(defun erc--server-propagate-failed-connection (process) + "Ensure the PROCESS sentinel runs at least once on early failure. +Act as a watchdog timer to force `erc-process-sentinel' and its +finalizers, like `erc-disconnected-hook', to run when PROCESS has +a status of `failed' after one second. Print the +`process-exit-status', which can be a number, like 111, or a +message, like \"TLS negotiation failed\"." + (when (eq (process-status process) 'failed) + (erc-display-message + nil 'error (process-buffer process) + (format "Process exit status: %S" (process-exit-status process))) + (pcase (process-exit-status process) + ((guard erc--server-reconnect-timer)) + (111 + (delete-process process)) + (`(file-error ,(rx "failed") ,(rx "unreachable") . ,_) + (erc-process-sentinel process "failed with code -523\n")) + ((rx "tls" (+ nonl) "failed") + (erc-process-sentinel process "failed with code -525\n"))))) + (defvar erc--server-connect-dumb-ipv6-regexp ;; Not for validation (gives false positives). (rx bot "[" (group (+ (any xdigit digit ":.")) (? "%" (+ alnum))) "]" eot)) @@ -708,9 +733,11 @@ erc-server-connect (with-current-buffer buffer (erc-current-nick)))) ;; wait with script loading until we receive a confirmation (first ;; MOTD line) - (if (eq (process-status process) 'connect) + (if (process-contact process :nowait) ;; waiting for a non-blocking connect - keep the user informed - (erc-display-message nil nil buffer "Opening connection..\n") + (progn + (erc-display-message nil nil buffer "Opening connection..\n") + (run-at-time 1 nil erc--server-connect-function process)) (message "%s...done" msg) (erc--register-connection)))) @@ -744,6 +771,68 @@ erc-server-delayed-reconnect (with-current-buffer buffer (erc-server-reconnect)))) +(defvar-local erc--server-reconnect-timeout nil) + +(defun erc-server-delayed-check-reconnect (buffer) + "Wait for internet connectivity before trying to reconnect. +BUFFER is the server buffer for the current connection." + ;; This may appear to hang for a good while at various places + ;; because it calls wait_reading_process_output a bunch. It does at + ;; least sometimes print "Waiting for socket from ..." or similar. + (with-current-buffer buffer + (setq erc--server-reconnect-timeout + (min 300 (* (or erc--server-reconnect-timeout + erc-server-reconnect-timeout) + 2))) + (let ((reschedule + (lambda (proc) + (with-current-buffer buffer + (let ((erc-server-reconnect-timeout + erc--server-reconnect-timeout)) + (delete-process proc) + (erc-display-message nil 'error buffer "Nobody home...") + (erc-schedule-reconnect buffer 0)))))) + (condition-case _ + (let ((proc (funcall erc-session-connector + "*erc-connectivity-check" nil + erc-session-server erc-session-port + :nowait t)) + (check-expire (time-add 10 (current-time))) + check-aside) + (setq check-aside + (run-at-time + 1 1 (lambda (proc) + (when (or (not (eq 'connect (process-status proc))) + (time-less-p check-expire (current-time))) + (cancel-timer check-aside)) + (when (time-less-p check-expire (current-time)) + (erc-display-message + nil 'error buffer "Timed out waiting on `connect'") + (delete-process proc) + (funcall reschedule proc)) + (when (eq 'failed (process-status proc)) + (funcall reschedule proc))) + proc)) + (set-process-filter + proc (lambda (proc _) + (delete-process proc) + (with-current-buffer buffer + (setq erc--server-reconnect-timeout nil)) + (run-at-time nil nil #'erc-server-delayed-reconnect + buffer))) + (set-process-sentinel + proc (lambda (proc event) + (with-current-buffer buffer + (pcase event + ("open\n" + (run-at-time + nil nil #'send-string proc + (format "PING %d\r\n" (time-convert nil 'integer)))) + ((or "connection broken by remote peer\n" + (rx bot "failed")) + (funcall reschedule proc))))))) + (file-error (funcall reschedule nil)))))) + (defun erc-server-filter-function (process string) "The process filter for the ERC server." (with-current-buffer (process-buffer process) -- 2.39.2