Ditto 2.3.0 is API and binary compatible to prior Eclipse Ditto 2.x versions.
Changelog
Eclipse Ditto 2.3.0 includes the following topics/enhancements:
- HTTP API for “live” commands
- Smart channel strategy for live/twin read access
- Configurable allowing creation of entities (policies/things) based on namespace and authenticated subjects
- Allow using
*as a placeholder for the feature id in selected fields
The following notable fixes are included:
- Fix potential concurrent modification errors when using JavaScript payload mapping and global variables
- Fix reconnect backoff for Kafka connections with authentication failures
- Fix caching of JWTs in HTTP push connections
- Fix potentially unreachable client actors in connections with
clientCount > 1 - Fix search inconsistencies for very active things during shard relocation (e.g. on rolling updates)
- Fix that a Kafka connection with only targets remains “open” even if Kafka broker is not available
- Allow usage of absolute domain paths ending with a “.” as Kafka bootstrap servers
- Ensure that Ditto pub/sub state is eventually consistent with a guaranteed upper time limit
The following non-functional work is also included:
- Update of several used dependencies:
- Pekko:
2.6.18 - Pekko Management:
1.1.2 - Caffeine:
3.0.5 - Classindex:
3.11 - Cloudevents:
2.3.0 - HiveMQ MQTT client:
1.3.0 - Jackson:
2.12.6 - Logback:
1.2.10 - SLF4J:
1.7.32 - SSL config core:
0.4.3 - Typesafe config:
1.4.1 - Rhino JS engine:
1.7.14
- Pekko:
For a complete list of all merged PRs, inspect the following milestones:
Compared to the latest release 2.2.0, the following most notable changes, new features and bugfixes were added.
New features
HTTP API for “live” commands
Ditto’s “live” channel is now also available for commands invoked via HTTP API.
See also the blogpost covering that topic.
To qualify a command (e.g. “modify thing” or “retrieve feature property”) as a “live” command, the header or
query parameter channel=live has to be specified.
Live commands bypass the twin and go directly to the devices.
However, an existing twin is still required for policy enforcement.
Smart channel strategy for live/twin read access
Ditto adds support for selecting the “twin” or the “live” channel for thing query commands based on an
RQL condition of a newly added parameter
live channel condition.
See also the blogpost covering that topic.
In addition, a new payload mapper automatically updating the twin based on received live data from devices was added.
Configurable allowing creation of entities based on namespace and authenticated subjects
This added feature allows configuring restrictions, which authenticated subjects may create new entities (things / policies) in which namespaces.
Allow using * as a placeholder for the feature id in selected fields
When selecting for certain fields of a thing or when using
signal enrichment (extraFields) in order to add more (unchanged) data from a twin to e.g. events
the wildcard * can now be used in order to select all features of a thing without the need to know their feature names.
Bugfixes
Several bugs in Ditto 2.2.x were fixed for 2.3.0.
This is a complete list of the
merged pull requests, including the fixed bugs.
Migration notes
No migrations required updating from Ditto 2.2.x
Ditto clients
For a complete list of all merged client PRs, inspect the following milestones:
Ditto Java SDK
No mentionable changes/enhancements/bugfixes.
Ditto JavaScript SDK
See separate Changelog of JS client.
Roadmap
Looking forward, the current plans for Ditto 2.4.0 are:
- Update service code to Java 17 (APIs stay at Java 8) + run Ditto containers with Java 17 runtime
- Continue work on the started WoT (Web of Things) integration