Should the NRF52Serial::setBaudrate() use a KeyValueTable to best approximate request?

[microbit-carlos]

Currently it's just a switch case where the users has to specify the exact baud rate from those available, or the defaul 115200 is applied instead:

• codal-nrf52/source/NRF52Serial.cpp#L193

int NRF52Serial::setBaudrate(uint32_t baudrate)
{
    nrf_uarte_baudrate_t baud = NRF_UARTE_BAUDRATE_115200;

    switch(baudrate)
    {
        case 9600 : baud = NRF_UARTE_BAUDRATE_9600; break;
        case 31250 : baud = NRF_UARTE_BAUDRATE_31250; break;
        case 38400 : baud = NRF_UARTE_BAUDRATE_38400; break;
        case 57600 : baud = NRF_UARTE_BAUDRATE_57600; break;
        case 115200 : baud = NRF_UARTE_BAUDRATE_115200; break;
        case 230400 : baud = NRF_UARTE_BAUDRATE_230400; break;
        case 921600 : baud = NRF_UARTE_BAUDRATE_921600; break;
        case 1000000 : baud = NRF_UARTE_BAUDRATE_1000000; break;
    }

    nrf_uarte_baudrate_set(p_uarte_, baud);
    return DEVICE_OK;
}

The micro:bit V1 Mbed implementation did something very similar to KeyValueTable, where best approximation below the request is applied instead:

• codal-core/source/core/CodalUtil.cpp#L33
• mbed-classic/targets/hal/TARGET_NORDIC/TARGET_MCU_NRF51822/serial_api.c#L108

void serial_baud(serial_t *obj, int baudrate)
{
    if (baudrate<=1200) {
        obj->uart->BAUDRATE = UART_BAUDRATE_BAUDRATE_Baud1200;
        return;
    }

    for (int i = 1; i<17; i++) {
        if (baudrate<acceptedSpeeds[i][0]) {
            obj->uart->BAUDRATE = acceptedSpeeds[i - 1][1];
            return;
        }
    }
    obj->uart->BAUDRATE = UART_BAUDRATE_BAUDRATE_Baud1M;
}

So this would align better with the V1 behaviour and also provide a baud rate closer to the one requested by the user (like DAL/CODAL do in other places like setting the acc/mag sampling rates, etc).

---

Somewhat related:

• Missing UART baud rates #285