- Add AVRO format support. Defaults to JSON
- Remove the Metadata repo configuration.
- Remove the Spring property sort-cuts. Only the native Debezium properties are used.
- change the native properties prefix from `cdc.config to cdc.debezium`.
- Removes the need to fork and modify the Debezium engine code.
- Add binding name strategy.
Remove old CDC projects and replace with the new implementation
Remove duplicate pom properties
remove non existing pom module
Rename `cdc-debezium-supplier/source` to `debezium-supplier/source`
streamline naming and tests
streamline supplier tests
test streamline. Github workflow
Fix Debezium engine generics mess
Break `DebeziumConfiguration` into engine auto-config and consumer configuration
add initial reactive consumer impl.
Splits Debezium into a pure supplier and a streaming source
- Remove any SCS related dependencies from the supplier.
- Supplier uses Sinks.Many.
complete the supplier vs source separation
Improve Debezium Egine auto-conf
- Provide configuration hooks for OffsetCommitPolicy, ConnectorCallback,
CompletionCallback and Clock with default implementations.
- Remove obsolete code.
minor improvements
Rename cdc property prefix to `debezium` and `cdc.debezium` to `debezium.inner`
remove obsolete `StreamBridge` tests
address reviews. improve testing
Supplier tests. Rename `inner` to `properties`.
Streamline the README. Minor improvements
streamline supplier tests
final fixes
Fixes https://github.com/spring-cloud/stream-applications/issues/431
* Use `WatchService` for local file `MessageSource` in `AwsS3SupplierConfiguration`
for `S3InboundFileSynchronizingMessageSource`.
This way we are able to scan sub-dirs in the local file store and skip those
sub-dirs as entities.
The `S3InboundFileSynchronizingMessageSource` is able to properly create a local sub-dir
according to the complex key of S3 object, for example if we have a remove entry as `myDir/myFile.txt`,
so in the `/local` dir it will be stored as `/local/myDir/myFile.txt`.
The mentioned `WatchService` can react to entries of entire file tree.
The script file may have several DDL expression to execute.
They probably have to be called one-by-one as they are declared.
For example our `init-db.cql` has DDL to delete table and the next to create.
* Fix `CassandraAppClusterConfiguration.keyspaceInitializer()`
to use a `Flux.concatMap()` instead of `flatMap()` to ensure sequential
items processing in the order they appear upstream
* WIP: initial xmpp supplier support
* implement xmpp supplier and test
* remove test logging config
* update for consistency
* add documentation references for xmpp consumer and supplier, fix formatting for functions where geode was commented out
