gpio:set_int(Gpio, Pin, none) -> error

[petermm]

Somebody struggled and posted this in the Discord. Reading the docs, it seems this code should work

Available triggers are none (which is the same as disabling an interrupt), rising, falling, both (rising or falling), low, and high.

https://www.atomvm.net/doc/main/apidocs/erlang/eavmlib/gpio.html#set-int-3

In my code I want to enable and disable interrupts but get an error when diabling the interrupts.

start() ->
    start(undefined, undefined).

start(_StartType, _StartArgs) ->
    Int_pin = 33,
    Gpio = gpio:open(),
    gpio:set_direction(Gpio, Int_pin, input),
    ok = gpio:set_int(Gpio, Int_pin, rising),
    ok = gpio:set_int(Gpio, Int_pin, none),
    wait_forever().

I get an error on the second gpio:set_int/3 call:

CRASH
======
pid: <0.1.0>

Stacktrace:
[{mpu6050_app,start,2,[{file,"/home/username/mpu6050/src/mpu6050_app.erl"},{line,21}]}]

cp: #CP<module: 4, label: 4, offset: 64>

x[0]: error
x[1]: {badmatch,error}
x[2]: {1,1,39,1,[{4,105}],error}

Problem solved by using:

ok = gpio:remove_int(Gpio, Int_pin)

But there may be a (doc?)/code issue.

[UncleGrumpy]

This looks like a bug, I will try to look into this over the weekend. You already found the workaround. Thanks for reporting this!

[UncleGrumpy]

I looked at our handling of gpio interrupts and we do not have a special case necessary to make none remove any interrupts set on the given pin. Since we have a dedicated function to disable the interrupts I think updating the docs makes more sense than making the code more complicated.

[UncleGrumpy]

I thought about this a bit more, and after thinking of all the situations where using the none trigger would be preferable, I looked harder and found that it could be supported with minimal changes. See #1399.