Update descriptor to increase range (16-bit values)?

When I connect the micro:bit to the computer, it recognizes it as a mouse and keyboard but I made the code click and i does not click the mouse.

Update name field to only be used in debug mode (maybe)?

May want to compare with Teensy Joystick (different usage page): https://github.com/PaulStoffregen/cores/blob/02b18c609abebb4b3565a9d4b8db43f3433c6f78/usb_hid/usb.c#L172

send_now(): https://github.com/PaulStoffregen/cores/blob/02b18c609abebb4b3565a9d4b8db43f3433c6f78/usb_hid/usb_api.cpp#L629

Init: https://github.com/PaulStoffregen/cores/blob/02b18c609abebb4b3565a9d4b8db43f3433c6f78/usb_hid/usb.c#L545

Report methods: https://github.com/PaulStoffregen/cores/blob/02b18c609abebb4b3565a9d4b8db43f3433c6f78/usb_hid/usb_api.h#L89

Two options:

• A pxt.json setting
• Dynamic memory (IFF Nordic stack can update to use new locations for VLOC user and variable length)

Review making the Report Characteristic const (in program space). It shouldn't change, but initial attempts (via just const) didn't seem to work.

Not sure if it's possible...See: https://github.com/T-vK/ESP32-BLE-Mouse

Add Gamepad support.

Some examples to draw from:

• https://gist.github.com/wereii/b215c799a267af10154087b5d5af3a6b
• https://stackoverflow.com/questions/41147796/hid-gamepad-report-descriptor-issue
• https://www.microchip.com/forums/m397098.aspx

Clean up branch mess (merge master into main / delete master; Merge in "Single Service" branch)

V1 build environment lacks the MicroBitBLEService class used as the base for HIDService.

Creating an alternative with the minimum features for HIDService may be sufficient (CreateService, CreateCharacteristic and utilize the BLE callbacks for on...() events).

Do a user-visible error when there's an attempt to use an uninitialized singleton.

How can you use spacebar? I had a look through the docs but I can't seem to find it.

Reduce memory: Replace static memory for report with table-based approach.

struct reporterEntry {
   uint8_t *data;
   uint16_t reportIDOffset;
   uint16_t length;
   reportEntry *next;
}

static reportEntry *reports;

Or maybe static array instead of linked list of nodes?

Notes

• Only supports input descriptors now

Misc. Finishing stuff.

• Testing
• Translations: Do to I support these everywhere?
• Review and clean-up memory use in general (make sure .ts stuff doesn't eat more mem than needed)
• Go to more standard debug stuff (DMESG) if it works...
• Should each service be a separate module/extension?
• Review / suggestions for semantics and affordances.
  • Mouse and Gamepad semantics are clunky. Maybe a smaller "dpad only" block for Gamepad and mouse x,y,button, hold for mouse?
  • Better use of types to enforce correct use.
• Get rid of setStatusChangeHandler and isEnabled() in each service. They were useful for development, but don't provide much benefit beyond Bluetooth connection blocks now. (This would also remove the Event ID concerns).
• Consider changing AbsMouse coordinates to be screen coordinates. (Upper left is (0,0))

V1

Ball is in someone else's court. This will require a pretty serious overhaul. See #7.

CODAL Changes

• HIDService needs memory for the characteristic that allow the report map to expand. The total size needs to be known before the service is created, which doesn't work well with the block semantics. Consequently a static buffer is created (reportMap) using the const reportMapMaxSize. Select a reasonable default value and have a pxt.json flag for larger values? (See #13)
• Advertising
  • Set a service UUID in the ad.
  • Set the appearance
  • Set the Pairing mode in the ad.
• Descriptors: Reports require special descriptors. It'd be nice if CODAL included support to add these as part of a characteristic.
• Event / Message ID stuff (too ad-hoc as-is) (the ID numbers in each Reporter are just made up / sequential.
• I've added a gross hack to monitor pm_events and restore CCCDs (which is already being done by the peer_manager, but not propagated to the MicroBitBLEChar's internal cccd field. This should be done as part of the peer_manager mechanics in MicroBitBLEManager instead. (See #15)
  • The MicroBitBLEChar's cccNotify and cccIndicate should be updated to use the CCCD value from the stack IF the characteristic has a CCCD. (Use sd_ble_gatts_value_get to get the value for the handle to the CCCD). Something like this for notify:

    if(handles.cccd == 0 || conn_handle==BAD) 
      return false;  
    uint16_t value; 
    ble_gatts_value_t data; 
    memset(&data, 0, sizeof(ble_gatts_value_t));
    data.len = 2; 
    data.p_value = &value; 
    sd_ble_gatts_value_get(conn_handle, handles.cccd, &data);  
    // Should this notify services of the update?  I think so... Use the onWrite method since it was a write to the value
    return data& BLE_GATT_HVX_NOTIFICATION;
  
  

  • If moved, then the pm_events and static_pm_events fields in HIDService.cpp can be removed.
  • See lancaster-university/codal-microbit-v2#160

Update documentation to describe the device name block and how it's related to the advertised name.

Need documentation:

• Use MakeCode README for block data.
• Maybe GitHub repo's wiki for devices that work/don't?
• Videos of use (how to pair / unpair)
  • Videos of cool applications

Currently events are limited to 100ms: minTimeBetweenNotifies

It looks like the max connection interval is 20ms (max_conn_interval ). That is, we should be able to increase it moderately (50ms?).

Need to do lots of testing. Some issues:

• Long-term runs
• Check AbsoluteMouse behavior
• Behavior on different platforms
• Combinations of services

macOS / iOS appear to "renew" notifications on connect, but Android and Windows don't. The peer manager should save/restore them, but it doesn't look like it's working.

Remove DMESG enable from pxt.json.

Error in MakeCode.

Here's details on the likely cause/workaround: https://forum.makecode.com/t/c-extensions-issues/6836/2

Move advertising to another module and make a dependency.
(And other services, like mouse, as separate modules...that can be composed)