Clarifications on the operation of the code and changing pins

[brightproject]

Good day and thanks for the repository @collin80 😄

At the very beginning I wanted to ask, why does the code use the can_common.h library if CAN bus is based on TWAI esp32?

I use the ESP32-S3 microcontroller + module SN65HVD230

The config from the .pio is as follows:

[env:ESP32_S3_BOARD]
platform =
espressif32
board = esp32-s3-devkitc-1-n16r8
framework = arduino

I have a custom board that I developed and there the MCU controls the encoder and display.
Of the free pins, only 17 and 18 are UART2 (PX2/TX2) by default
The file esp32_can.cpp contains the default code.
From the code example

esp32_can/examples/CANTestESP32_BuiltIn_AltPins/CANTestESP32_BuiltIn_AltPins.ino

Line 28 in 66d1440

pinMode(GPIO_NUM_16, OUTPUT);

it is clear that pin 16 needs to be pull to ground and then other contacts reassigned.
My code:

#include <esp32_can.h>

void setup() {
  Serial0.begin(115200);

  Serial0.println("Initializing");

  pinMode(GPIO_NUM_16, OUTPUT);
  digitalWrite(GPIO_NUM_16, LOW); //enable CAN transceiver
  CAN0.setCANPins(GPIO_NUM_17, GPIO_NUM_18);
  CAN0.begin(500000);

  Serial0.println("Ready");

  CAN_FRAME txFrame;
  txFrame.rtr = 0;
  txFrame.id = 0x123;
  txFrame.extended = false;
  txFrame.length = 5;
  txFrame.data.uint8[0] = 0x1A;
  txFrame.data.uint8[1] = 0x2B;
  txFrame.data.uint8[2] = 0x3C;
  txFrame.data.uint8[4] = 0x4D;
  txFrame.data.uint8[5] = 0x5E;
  CAN0.sendFrame(txFrame);

}

void loop() {

}

This code is executed once and at the end of the Error frame.

If I change the pins in the esp32_can.cpp file forcibly to

ESP32CAN attribute((weak)) CAN0(GPIO_NUM_17, GPIO_NUM_18) ;

Or I write them in the main code in the SETUP section

void setup() {
  Serial0.begin(115200);

  Serial0.println("Initializing");

  CAN0.setCANPins(GPIO_NUM_17, GPIO_NUM_18);
  CAN0.begin(500000);

  Serial0.println("Ready");

  CAN_FRAME txFrame;
  txFrame.rtr = 0;
  txFrame.id = 0x123;
  txFrame.extended = false;
  txFrame.length = 5;
  txFrame.data.uint8[0] = 0x1A;
  txFrame.data.uint8[1] = 0x2B;
  txFrame.data.uint8[2] = 0x3C;
  txFrame.data.uint8[4] = 0x4D;
  txFrame.data.uint8[5] = 0x5E;
  CAN0.sendFrame(txFrame);

}

then the frames will start to be sent continuously with breaks between sendings of 54 us and each frame ends with a NAK(didn't fit in the screenshot).

What am I doing wrong or right?
Why does the pin reassignment example recommend doing it this way, and sending only once

  pinMode(GPIO_NUM_16, OUTPUT);
  digitalWrite(GPIO_NUM_16, LOW); //enable CAN transceiver
  CAN0.setCANPins(GPIO_NUM_17, GPIO_NUM_18);

But my code worked like this and sending is continuous
CAN0.setCANPins(GPIO_NUM_17, GPIO_NUM_18);
I would like to understand this issue.