node-red-contrib-ccu 4.7.0

v4.7.0

on Node.js NPM

Added

• On openccu-lite the interfaces are reached through the system's lite-rpc
  (with occulite-client)
  instead of XML-RPC/BIN-RPC clients and callback servers of this node's own -
  the requests over the system's web port, the events from a stream the
  connection opens, replayed after a break and re-read after a restart of the
  system; one subscriber at the interface processes serves every program on
  the system, and nothing registers with init. The credential is the
  openccu-lite token of the connection node, or on the system itself
  RedMatic's own token, which the system mints for it. A new setting
  openccu-lite interfaces in the connection node chooses: auto (the
  default; the CCU way on a CCU), openccu-lite (lite-rpc only) or CCU (the
  direct connection with callbacks, on openccu-lite too). An existing
  connection is moved to lite-rpc once, at its first start on openccu-lite,
  and the log says so; a choice made in the node is never changed by a later
  version. The value cache is filled from the system's state store at deploy,
  so nodes with cache get a value on openccu-lite as well. CUxD and CCU-Jack
  keep their own connections. On a CCU, RaspberryMatic and OpenCCU nothing
  changes.

Changed

• An interface process that is not up yet is tried every 3 s for the first
  90 s, then every 15 s. The 1, 2, 4, 8 s steps and the 15 s after them
  missed hmipserver's ready moment on a booting system by up to 15 s (measured
  on an openccu-lite Pi 4: hmipserver answers 50-55 s after the boot, the next
  attempt came at 61 s). The same holds after a process went away. When one
  interface appears, the others still waiting are tried at once, and the log
  line that reports the connection says how long after the start (or after the
  outage) it came. The metadata API's re-detection on openccu-lite follows the
  same schedule.
• The port of each interface can be set in the connection node, beside its
  checkbox: blank is the CCU's port for the transport as before (2001/42001 and
  so on, the 3xxxx ports on the CCU itself), a number replaces it - for a CCU
  behind port forwarding, or a simulator on another port.
• The test suite runs hm-simulator 1.4 on free ports, so it runs beside a CCU
  or another suite, and drives the simulator's scenario API: devices paired
  and removed while Node-RED runs, a ramp-time write, a service message, an
  interface process that restarts and forgets its clients, TLS with basic
  auth.
• On openccu-lite the metadata stream reports as Metadata, not as
  "Interface ReGaHSS". The system has no ReGaHSS; the connection borrowed
  that slot for the stream's state, so the log said "Interface ReGaHSS
  connected" at every start and reconnect, right next to "this box has no
  ReGaHSS". The log now says "Metadata stream connected", the nodes' status
  counts the stream under that name, and the editors' interface lists do not
  offer it.

Commits since v4.6.2

• 4.7.0 - openccu-lite through lite-rpc by default, interface ports, 3 s retries (f5a4e5b)
• docs(changelog): 4.7.0 dated (d288eb6)
• fix(lite): a token with a higher rpc tier is not taken for names-only (8256d81)
• feat: on openccu-lite the interfaces go through the system's lite-rpc with occulite-client (18329ca)
• test: the init retry spec expects the 3 s retry lines with the time since the start (483b75b)
• test: the suite on hm-simulator 1.4 with free ports and its scenario API; interface ports in the connection node (f3d0e5e)
• fix(meta): the metadata stream reports as Metadata, not as Interface ReGaHSS (a1af0e7)
• docs: the credentials migration came with 4.5.0, not 4.4.6 (49697ff)
• feat(rpc): try a missing interface process every 3 s for 90 s, then every 15 s (e7b1347)