introduce process:died? to make backoff logic more readable

pppoe.fnl

@@ -18,8 +18,7 @@
                   ipstate-script
                   ipstate-script))]
     (each [event (event.next-event transport-device pppd)]
-      (when (not pppd.running?)
-        (pppd:backoff))
+      (when (pppd:died?) (pppd:backoff))
       (when (and (not pppd.running?)
                  (netdev.link-up? transport-device)
                  (pppd:backoff-expired?))

tests/pppoe-test.fnl

@@ -49,6 +49,7 @@
      (when (= nil p.backoff-until)
        (set p.backoff-until (+ (clock) p.backoff-interval))
        (set p.backoff-interval (* 2 p.backoff-interval))))
+   "died?" (fn [p] (and (not p.running?) (= p.backoff-until nil)))
    "backoff-expired?"
    (fn [p]
      (and p.backoff-until (<= p.backoff-until (clock))))
@@ -82,10 +83,9 @@
             (var started false)
             (set my-events [1 2 3 4 5 6 7 8 ])
             (mock :process :join #(set joined true))
-            (tset p "start" #(set started true))
             (pppoe "eth0" "ppp0")
             (assert joined "ifconfig process did not join")
-            (assert started "daemon did not start")))
+            (assert p.running? "daemon did not start")))
 
         (lambda backoff-increases-on-failure []
           (var delay 0)