Unhashable type for Systec interface errors

[Chairless-maybe]

Describe the bug

When running DroneCAN on top of the systec interface, error messages that appear raise TypeError: unhashable type exceptions. This swallows the actual error messages and prevents troubleshooting.

The cause of error messages is likely a config/hardware issue, so it is probably irrelevant. What matters is that the CAN error messages disappear and are not relayed to the user.

To Reproduce

Below is a script to monitor messages coming through DroneCAN. Error messages do not always happen, but they tend to be triggered by sending commands through DroneCAN.

The actual mechanism triggering error messages is more likely a config/hardware issue than anything.

Expected behavior

The error message raised during operation should be exposed to the user.

Additional context

OS and version: Windows 11
Python version: 3.10
python-can version:
python-can interface/s (if applicable): systec

Traceback and logs

Traceback:

Traceback (most recent call last):
  File "c:\Users\thoma\OneDrive\Documents\Tyto\Python scripts\test.py", line 51, in <module>
    dc_node.spin(timeout=0.001)
  File "C:\Python310\lib\site-packages\dronecan\node.py", line 439, in spin
    execute_once()
  File "C:\Python310\lib\site-packages\dronecan\node.py", line 431, in execute_once
    frame = self._can_driver.receive(read_timeout)
  File "C:\Python310\lib\site-packages\dronecan\driver\python_can.py", line 137, in receive
    self._check_write_feedback()
  File "C:\Python310\lib\site-packages\dronecan\driver\python_can.py", line 124, in _check_write_feedback
    raise item
  File "C:\Python310\lib\site-packages\dronecan\driver\python_can.py", line 101, in _writer_thread_loop
    self._bus.send(msg)
  File "C:\Python310\lib\site-packages\can\interfaces\systec\ucanbus.py", line 212, in send
    self._ucan.write_can_msg(self.channel, [message])
  File "C:\Python310\lib\site-packages\can\interfaces\systec\ucan.py", line 488, in write_can_msg
    UcanWriteCanMsgEx(self._handle, channel, c_can_msg, c_count)
  File "C:\Python310\lib\site-packages\can\interfaces\systec\ucan.py", line 109, in check_result
    raise UcanError(result, func, arguments)
  File "C:\Python310\lib\site-packages\can\interfaces\systec\exceptions.py", line 16, in __init__
    message = self._error_message_mapping.get(result, "unknown")
TypeError: unhashable type

Script to recreate the bug:

import can
import dronecan
from dronecan.driver import python_can as pycan_driver
from dronecan.app.node_monitor import NodeMonitor
import time

can_bus = pycan_driver.PythonCAN(channel=0, interface="systec", bustype="systec", bitrate=1000000)

node_id = 45

dc_node = dronecan.node.Node(can_bus, node_id=node_id, bitrate=1000000)

print("=== Local Node ===")
print(f"  node_id : {dc_node.node_id}")
print(f"  mode    : {dc_node.mode}")
print(f"  health  : {dc_node.health}")
print()

# Print every NodeStatus as it is received:

def on_node_status(event):

    print(
        f"[NodeStatus] from node {event.transfer.source_node_id}: "
        f"uptime={event.message.uptime_sec}s "
        f"health={event.message.health} "
        f"mode={event.message.mode}"
    )
 
dc_node.add_handler(dronecan.uavcan.protocol.NodeStatus, on_node_status)

LISTEN_SECONDS = 10

monitor = NodeMonitor(dc_node)

print(f"Listening for {LISTEN_SECONDS} seconds...\n")
deadline = time.monotonic() + LISTEN_SECONDS
while time.monotonic() < deadline:
    try:
        dc_node.spin(timeout=0.001)
        msg = dronecan.uavcan.equipment.esc.RPMCommand()
        msg.cmd = int(0)
        dc_node.broadcast(msg)
    except dronecan.transport.TransferError as ex:
        print(f"  (transfer error, continuing: {ex})")

dc_node.close()

With Claude, I tracked down the error to interfaces/systec/exceptions.py. A ctypes.c_ubyte error code is passed to the Python interface, and the systec backend fails to interpret it. By changing Line 16 in exceptions.py, I can see the underlying exception and original error message:

message = self._error_message_mapping.get(result, "unknown")

should become

message = self._error_message_mapping.get(result.value, "unknown")

This seems to have solved the issue, and now I can troubleshoot my hardware problems better. I cannot make a PR for this at the moment, but I might later.

[namoetheintern]

I can't be of any help because I'm having debugging issues, too, but do you mind telling me what hardware connections you made on your pi hat and what you're interfacing with?