@bitbank2 I opened this to trigger the CI; would you mind testing it in your environment before we merge it? Thanks

I'll give it a try today. Thanks :)

I have two 2 tests for the BLE speed. One connects to a BLE thermal printer and prints some graphics and the other uses my VirtualDisplay and sends SSD1306 gfx data over the BLE connection. The Thermal printer test shows that your fix works, but the VirtualDisplay doesn't show any speed improvement. The ESP32 is able to send data 5x faster than the Nano33 under the same conditions. I don't understand why the data transmission on the Nano33 is so much slower even when writeWithoutResponse is used. The last comment I have is that the option to write without response should probably be added to all of the various writeValue() methods too.

Hi@bitbank2 , so this patch is working and it causes the same effects of#72. I guess that the speed problems also exist with your other solution, am I correct?
Then, the BLE speed depends on:

• the connection interval range, set during devices connection (here). The smaller the number, the faster it is. The minimum is already set to the minimum allowed: 0x0006 (corresponding to 1.25 ms). You could try to force the actual connection interval to the minimum by setting also the maximum value to 0x0006 this way:
  if (HCI.leCreateConn(0x0060, 0x0030, 0x00, peerBdaddrType, peerBdaddr, 0x00, 0x0006, 0x0006, 0x0000, 0x00c8, 0x0004, 0x0006) != 0) {

• the actual BLE controller, that in case of the nano33 is CORDIO from arm. You can try with a nano iot that has a different controller.

Please merge and close this PR.