read_pulses() still blocks

[mfeif]

I've read through #18 after having trouble reading repeat commands from my IR remote. Here's some code to illustrate what I'm seeing, which is that while the remote is "clicked" on anything, the loop gets stuck:

(this is a light change from the bundled example file in this repo)

import pulseio
import board
import adafruit_irremote

pin = board.D17

pulsein = pulseio.PulseIn(pin, maxlen=120, idle_state=True)
decoder = adafruit_irremote.GenericDecode()

while True:
    pulses = decoder.read_pulses(pulsein, blocking=False)
    if pulses:
        print("\nHeard", len(pulses), "Pulses:", pulses, flush=True)
        try:
            code = decoder.decode_bits(pulses)
            print("Decoded:", code)
        except adafruit_irremote.IRNECRepeatException:
            pass
        except adafruit_irremote.IRDecodeException as e:
            pass
    else:
        print('.', end='', flush=True)

If you run this, you'll see a flood of "." filling the console, as expected, but as soon as you put some activity on the pin, the flood will pause. This will pause indefinitely... I held down for 10+ seconds, and it blocked. When I let go, only one event was returned.

I tried to reason through _read_pulses_non_blocking() in adafruit_irremote.py but I'm having trouble, since I don't really know what it's trying to do, so I can't spot the bug in the logic there.

I've also seen Kevin's work here that includes a timeout feature that seems to taking a stab at the same thing. It works, and even supports the repeat exception that in the example code.

[mfeif]

I now have an ItsyBitsy M4, and I'm seeing the same thing as I saw above on my rPi.

It still blocks while a button is pressed down.

[majuss]

Same problem, sadly this lib is unusable in the current state.

[danielballan]

I can reproduce this. A continuous stream of pulses (caused by holding down a button the remote) causes the read_pulses to block even when blocking=False is set. I modified the irremote_simple.py to add timing output as follows

$ diff examples/irremote_simpletest.py /media/dallan/CIRCUITPY/code.py 
8a9
> import time
15c16,22
<     pulses = decoder.read_pulses(pulsein)
---
>     t0 = time.monotonic()
>     pulses = decoder.read_pulses(pulsein, blocking=False)
>     t1 = time.monotonic()
>     if t1 - t0 > 0.02:  # 20 ms
>         print(f"Blocked for {t1 - t0:.2} seconds")
>     if pulses is None:
>         continue

The output from a short button press looks OK:

code.py output:
Blocked for 0.4 seconds
Heard 41 Pulses: [2422, 579, 1224, 569, 626, 568, 1224, 568, 1224, 568, 1225, 567, 1225, 568, 627, 567, 628, 566, 1226, 567, 628, 566, 1226, 566, 1226, 567, 628, 566, 1226, 566, 629, 566, 629, 565, 630, 564, 1228, 565, 1227, 565, 1227]
Decoded: [67, 75, 8]
----------------------------

but the output from holding down the button reproduces the issue described above.

Blocked for 7.8 seconds
Heard 41 Pulses: [2434, 568, 1223, 570, 625, 569, 1234, 559, 1233, 559, 1233, 560, 1233, 560, 634, 560, 625, 569, 1234, 559, 636, 559, 1233, 559, 1233, 560, 635, 560, 1232, 561, 634, 560, 635, 560, 625, 569, 1223, 569, 1233, 559, 1233]
Decoded: [67, 75, 8]
----------------------------

I have an idea of how to fix this. PR forthcoming...

[danielballan]

My fix works well. See #42.

[danielballan]

I think it safe to close this now that #42 is in.

[tannewt]

Thanks @danielballan!