'window' input mode doesn't use hysteresis on change

[dndrks]

repro script (confirmed both with self-patching output 1 to input 2 and sending from another module):

-- self-patch output 1 to input 2
function init()
  input[2].mode('window',{-2,1,2,4},0.1)
end

input[2].window = function(c,d)
  if d then
    print('hello')
  else
    print('goodbye')
  end
end

when we hit any of the window values (eg. 2V), there's a deluge of salutations/valedictions printed. i assumed that the hysteresis argument would buffer against this, but perhaps i'm misunderstanding its function? <333

[xmacex]

I tried the above script via druid, yes I get the same behavior. I put a little counter inside the handler before the if statement, and that is unstable. Many events are happening during physically plugging or unplugging the jack, something like 5-20 but once a few dozen. Not as many salutations/valedictions¹ get printed.

Newest crow firmware and druid 1.1.5 running on norns, I reset my crow before trying.

¹ I learned a word today.

[trentgill]

this makes sense. looking at the detection code, there is no hysteresis being applied. if some enterprising individual wants to look at it, see here:

crow/lib/detect.c

Line 272 in 4ca8b7d

static void d_window( Detect_t* self, float level )