How to detect when a function was activated by a timer?

Question

When running a function inside a timer, there are some constraints on whats possible.

Adding timers should not be done (according to the documentation).
Communicating with a sub-process is limited (see Calling flyspell from a timer & Blocking call to accept-process-output with quit inhibited!).

Is it possible to detect if a function is running from a timer so an alternate code-path can be used?

The only solution I've found is to add advice on the caller to define a variable which the function being called can check, however this isn't such a nice solution as it may mean adding advice to many functions.

I guess you can see it from the call stack, which you can loop through with `mapbacktrace`. — Gilles 'SO- stop being evil', Sep 05 '22 at 12:05
@Gilles'SO-stopbeingevil' : Is that viable at runtime? How does a (non-debugger) Lisp program go about obtaining the current call stack? — Phil Hudson, Sep 07 '22 at 13:26
@PhilHudson Lisp programs don't need any official “debugger” blessing to call `mapbacktrace`. It will miss frames that are optimized away by the byte compiler, but that wouldn't matter here. — Gilles 'SO- stop being evil', Sep 07 '22 at 17:34

score 2 · Answer 1 · answered Sep 05 '22 at 10:50

All timers get processed in timer-event-handler, so you can add advice to that instead:

(defvar called-by-timer nil
  "Bound to t when a function is called by a timer.")

(advice-add 'timer-event-handler :around
            (lambda (oldfun &rest args)
              (let ((called-by-timer t))
                (apply oldfun args))))

(defun hello ()
  (message "%s %s" "Hello" (if called-by-timer "from timer" "from elsewhere")))

(hello)
(run-with-timer 1 nil #'hello)

Should `hello` not also unset `called-by-timer` on exit? – Phil Hudson Sep 07 '22 at 13:21 — Phil Hudson, Sep 07 '22 at 13:21

How to detect when a function was activated by a timer?

1 Answers1