From dd173738c8ba33c16a0ca4a2e2aa0298fec8525e Mon Sep 17 00:00:00 2001
From: Mattias <30832155+bammab@users.noreply.github.com>
Date: Sat, 26 Oct 2024 16:16:56 +0200
Subject: [PATCH] Fix reconnect

If the watchdog trigger connection_lost and the serial connection if already
open, it is not possible to close/abort the connection.
Old:
- Watchdog timout
- connection_lost (transport set to None)
- _reconnect
  - _disconnect (which could not abort, because transport not set anymore)
  - _create_connection
    Call pyserial-asyncio: create_serial_connection -> _ensure_reader -> _read_ready -> _close -> connection_lost (end here, because of lock)

New workflow could abort the transport and the new connection will work.
Anyway connection_lost will be called twice:
1. by the watchdog (with TimeoutError)
2. by _reconnect -> _disconnect -> abort
The last one will do nothing more, because of the lock of _reconnect
---
 sml/asyncio.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/sml/asyncio.py b/sml/asyncio.py
index c721fe6..fb96c25 100644
--- a/sml/asyncio.py
+++ b/sml/asyncio.py
@@ -86,7 +86,6 @@ def data_received(self, data: bytes):
 
     def connection_lost(self, exc: Optional[Exception]):
         logger.debug('port closed')
-        self._transport = None
         if self._running and not self._lock.locked():
             asyncio.ensure_future(self._reconnect(), loop=self._loop)