Serial mysteries

[boisgera]

When I am using println in Go on the Arduino, the Python serial code I am using "works" but:

• I don't lose any data, even if I start the reading 10 sec after the flash. Is it some buffering effect on the computer or at the arduino level? (Given the LED on the Arduino, I'd say that there is no data send on Serial)
• When I restart the Python program, the Arduino programs is restarted !

tinygo monitor works differently ; some data is lost if I delay the read, then only a few displays are performed, and the second try may not print anything at all.

I'd like to understand better what's going on here

[boisgera]

Ref: https://github.com/tinygo-org/tinygo/blob/b5ad81c884981f42c1ac491cc1d195aea886ddbe/monitor.go

TODO: Study the influence of baudrate.

[boisgera]

OK, with the proper baudrate specified,

tinygo monitor -baudrate 9600

the behavior of tinygo monitor is not identitcal to the Python code

import serial

BAUD_RATE = 9600
SERIAL_PORT = "/dev/ttyACM0"

with serial.Serial(SERIAL_PORT, BAUD_RATE) as file:
    while True:
        bytes = file.readline().strip()
        print(bytes.decode("utf-8"))

in other words, starting the monitoring of tinygo monitor restarts the program. Let's say that the witness program is

package main

import "time"

func main() {
	for i := 0; ; i++ {
		println("Hello!", i)
		time.Sleep(time.Second)
	}
}

I'll end up with

Hello! 0
Hello! 1
Hello! 2
Hello! 3
Hello! 4
...

everytime, except that with tinygo monitor, I'll typically end up with the buffer clearing first (?)

$ tinygo monitor -baudrate 9600
Connected to /dev/ttyACM0. Press Ctrl-C to exit.
H! 45
Hello! 46
Hello! 47
Hello! 48
Hello! 49
Hello! 50
Hello! 0
Hello! 1
Hello! 2

So I know that the program on the Arduino keeps running and is not blocked because the write on the Serial port would be blocked. Some data that is not read is also forgotten, so I guess that the Serial file has a cyclic buffer or something similar (note: that may also explain why the very first read is also a bit borked). Now how come the Python program does not pick this buffer? Note that a variant of the Python program with a lower-level read (byte-by-byte) also works "perfectly" (i.e. does not print the data that may have been buffered but starts with 0 every time).

OK, the reset of the program with an opening of the connection is intented. See https://forum.arduino.cc/t/does-arduino-reset-when-first-connected-to-usb/544278/8:

The Arduino is reset by opening the serial port (auto reset) and/or on power up. You can prevent the auto reset from the serial port, temporarily, by connecting a 10uf cap between the reset pin and ground. If using an electrolytic cap, make sure to observe polarity. Some Arduinos have a jumper that you can cut to permanently disable the auto reset.

OK, we can even use the RESET pin to restart programmatically, etc. That could be useful if there is a panic, that kind of stuff. Now the only (slight) mistery remaining is how the tinygo monitor acces the Serial port buffer and Python pyserial does not. But this is not a critical issue ATM. Maybe it's only a temporisation that takes place in the Python code but not in the tinygo code.

[boisgera]

TODO: test tinygo flash -monitor (cf. https://twitter.com/sago35tk/status/1606454421273350144?s=20&t=fVk6i3E6rf0FrRm9hZszVA)