i2c write fail incorrect

[Deous]

This part of code makes some identical oleds not to work

	if (!i2c->write(address)) {
		i2c->stop();
		ESP_LOGE("oled", "OLED I2C bus not responding.");
		goto oled_init_fail;
	}

It is caused by I2C::write()

// return: true - ACK; false - NACK
bool I2C::write(uint8_t data)
{
    uint8_t ibit;
    bool ret;

    for (ibit = 0; ibit < 8; ++ibit)
    {
        if (data & 0x80)
            _SDA1;
        else
            _SDA0;
        _DELAY;
        _SCL1;
        _DELAY;
        data = data << 1;
        _SCL0;
    }
    _SDA1;
    _DELAY;
    _SCL1;
    _DELAY;
    ret = (_SDAX == 0);  // it is always 0 on some oleds
    _SCL0;
    _DELAY;

    return ret;
}

If we ignore __SDAX==0 and return true all works fine. Might related be some incorrect protocol response

[Yi Xie]

According to the datasheet of SSD1306, ACK (SDAout=0) should be sent after every byte. I guess some of the panels only connects SDAin without SDAout to SDA pin. In this situation of cause ACK should be ignored. However I don't think it's a good idea to always ignore ACK no matter whether SDAout is connected. For details please refer to pp.19 of the datasheet.

[Deous]

The interesting thing is - those oleds somehow work fine on Arduino libraries - particularly Adafruit.
I haven't looked deeper there because it is heavy wrapped but they do some kind of oled test before switching it on. Maybe something should be done in our case