Also need to check D21.toggle()

This fails.

Works fine on my board with:

• D21.set() --> led on
• D21.reset() --> led off
• D21.toggle() --> led on
• D21.toggle() --> led off
  No idea whats going wrong on your board.
  Only to make sure, compiled from ESP32 branch

set & reset works for me too.

Seems to be issue with:

digitalPulse(D22,0,10);

I see a quick flash.

I believe it's meant to leave the LED lit.

Just tested toggle one more time directly after reset without doing any command.

• --> led is off
  

• D21.toggle() --> led stays off
• D21.toggle() --> led on
• D21.toggle() --> led off
• D21.toggle() --> led on
  So first toggle only has a problem, if pin is not used before.
  Will check digitalPulse later today

I get different output! 1v91.390
Toggle only works after a reset()!

Then won't toggle off.

For the pulse:

https://github.com/espruino/Espruino/blob/master/targets/esp32/jshardware.c#L396
int duration = (int)pulseTime * 1000; //from millisecs to microsecs

https://github.com/espruino/Espruino/blob/master/targets/esp32/jshardwarePulse.c#L92

//pin to be pulsed. value to be pulsed into the pin. duration in milliseconds to hold the pin.
void sendPulse(Pin pin, bool pulsePolarity, int duration){

Could be a factor of 1000 out - comments don't match up.

Correct, comments don't match. But thats not the problem.
Looks like duration is not used. There is only one pulse.
So one pulse is working fine.
Have to find a nice way to stop.

Oh, sorry, misunderstanding of duration. I have to check the us / ms now.

Tested with digitalPulse(D21,0,[4,5,6,7,8]); --> works fine
next test with digitalPulse(D21,0,[4,50,6,7,8]); --> 2nd pulse (50ms) is only 17 ms long

RMT timer has a base frequency of 80Mhz.
This can be divided by 1 ... 255
Duration of pulse is 1... 32768
So longest pulse time would be 104ms, at least in theory.
Some more investigation is needed now from my side.
Actually, a divider of 80 is used in Espruino, therefore we use recalculation to usec

Found the reason, why D21.reset() did not work.
Config of rmt connects a pin to rmt directly.
Due to that, GPIO cannot set/reset D21 anymore, its not connect to the pin anymore.
According to espressif we would need a gpio_matrix_out with SIG_GPIO_OUT_IDX
We could do this after each pulse, it would take some CPU time.
Any comments to this ?

Can we set the config for RMT only to occur when digitalPulse is used?

I wonder of this is also making Onewire inconsistent?

The only way, I know, is to config rmt before pulse and to reset to GPIO after the pulse.
Have to find a way to recognize end of pulse (?)
Or the other way, we could switch to GPIO, whenever we set/reset/toggle a pin.

Or the other way, we could switch to GPIO, whenever we set/reset/toggle a pin.

That might the cleanest way

Is done in ESP32 branch.
Had to reset to rmt in jshardwarePulse too.
Hopefully, this did not create conflicts somewhere else (SPI or I2C for example).
Same problem could appear, if you PIN.set() for an already for SPI assigned pin

@jumjum123

Is this all sorted now - can we close, or does it need a pull request to bring into master?

@jumjum123

Can we close this now?

Works as expected,
as long as nobody finds something different ;-)