node-red-contrib-ccu 4.4.1

v4.4.1

on Node.js NPM

Fixed

• Node-RED could be killed by a failed ReGaHSS write, seconds after
  start (RedMatic #601).
  A write to a system variable that arrives before the variable list is
  known is deferred into a queue and flushed once the first
  getRegaVariables completes. That flush dropped the resulting promise,
  so a rejected write — ReGaHSS closing the connection, Error: socket hang up — became an unhandled rejection, which Node >= 15 turns
  into an uncaught exception: the whole Node-RED process exited with
  status 1 and its supervisor restarted it into the same crash. The
  rejection is now caught and logged.

  This became reachable in 4.3.0: the rewritten local-CCU detection made
  isLocal true again on RedMatic boxes (it had silently stopped
  matching on current firmware), which moves ReGa traffic from lighttpd
  on 8181 to ReGaHSS directly on 8183. lighttpd absorbed a busy ReGaHSS;
  the direct port simply closes the connection.

• The same floating-promise pattern in ccu-mqtt: setVariable,
  programActive and programExecute triggered over MQTT dropped their
  promises too, and now report through the node's error output instead of
  taking the process down.

• regaPoll's finally chain is caught as well, so a throw in the
  follow-up scheduling cannot escape as an unhandled rejection.

Commits since v4.4.0

• 4.4.1 (6689bdc)
• HANDOFF: 4.4.0 released (B-17, openccu-lite) (e134ebd)