Memory leak with idle checking

[jacwright]

When idle checking is on, every mouse move, keystroke, and scroll event calls wake which iterates through an array of old timers and clears them (though only effectively clearing the last timer which would be the only one ever active). This array grows unbounded with each event. An array should not be used here.

Current:

trackIdleStatus = ->
    timer = []
    wakeUp = ->
      timer.map(clearTimeout);
      ifvisible.wakeup()  if status isnt "active"
      idleStartedTime = +(new Date())
      timer.push setTimeout(->
        ifvisible.idle()  if status is "active"
      , idleTime)
...

Fix:

trackIdleStatus = ->
    timer = null # No array, just a reference to the last timeout id
    wakeUp = ->
      clearTimeout(timer); # Clear previous timeout. A noop if the timeout already ran
      ifvisible.wakeup()  if status isnt "active"
      idleStartedTime = +(new Date())
      timer = setTimeout(->
        ifvisible.idle()  if status is "active"
      , idleTime)
...

[AbelMakihara]

When page is idle and mousemove to 'wake up', ifvisible.wakeUp(wakeUp) and 'mousemove' clear the same timer and set 2 timer.
So one of the timers won't be cleared and it will always trigger 'idle' callback.
That's what i found in v1.0.6. :(

[jacwright]

@AbelMakihara good catch. Here is an updated fix:

Fix:

trackIdleStatus = ->
    timer = null # No array, just a reference to the last timeout id
    wakeUp = ->
      return ifvisible.wakeup()  if status isnt "active"
      clearTimeout(timer); # Clear previous timeout. A noop if the timeout already ran
      idleStartedTime = +(new Date())
      timer = setTimeout(->
        ifvisible.idle()  if status is "active"
      , idleTime)
...

[woowalker]

is still on maintain ?

[rosskevin]

I do believe this to be fixed in my fork https://github.com/rosskevin/ifvisible. If not please PR a test (expanded tests are present there)