v0.18d

Changes since V0.18c:

-- show default for mqtt.clientid in addon config dialog

Changes since V0.18b:

-- added mqtt.clientid to addon config dialog

Changes since V0.18a:

-- added mqtt.username/password to addon config dialog

Changes since V0.18:

-- update addon VERSION

Changes since V0.16:

-- support MQTT authentication
-- fix topics pushed in response to /get/ request

v0.16

Changes since previous release (v0.12):

0.16 - 2015/08/12 - owagner

retry a xml-rpc request at most once (after a 10s grace period). Previously, would retry requests indefinitely with a 30s delay, which defeated the ping mechanism

0.15 - 2015/07/26 - owagner

accept PING reply events (events with address "CENTRAL"), and use "ping" calls as a first-step measure in case of idle timeouts. Re-Init will still be sent after twice the specified timeout value. Change default idle timeout to 120s.

0.14 - 2015/07/25 - owagner

new option "hm.jsonNameTable" to specify a file which includes a JSON encoded table with device and channel address to name mappings (e.g. generated by hm-manager)

0.13 - 2015/07/19 - owagner

updated to eclipse-paho 1.0.2 and minimal-json 0.9.2
ignore intermediate syslog I/O errors

Timestamps, no more automatic reportValueUsage, more meta-information, new option hm.bindaddress

Merge remote-tracking branch 'origin/master'

Conflicts:
	.gitignore

CCU2 addon startup fix when CCU2 hostname was not resolveable

v0.10

- REALLY do not call getLocalHost()

New internal device management

• completely reworked internal device management. Now properly implements the
  listDevices/newDevices/deleteDevices contract recommended by HM's XML-RPC API, and will store local
  device information in a file (default: hm2mqtt.devcache)
• now correctly calls reportValueUsage with count=1 on every channel's datapoint. Note that this may
  result in a large number of pending configurations if devices weren't previously strongly linked to
  the XML-RPC server
• will now attempt to refetch ReGa names every 10 minutes everytime a name fails to resolve.
  When a ReGa fetch worked once (i.e. a ReGa is known to be present), failed names will always be
  published with the retain flag set to false, to avoid the unresolved names ending up in the
  MQTT broker's persistent storage