Kamu

Changed
- `InitiatorFilterInput` now accept `[AccountID]` instead of `AccountName`
`AccountFlowFilters` now filter by `DatasetId` instead of `DatasetName`.
- Upgraded to `datafusion v38`
Added
- Added a public image with [sqlx-cli](/images/sqlx-cli)
- Added a [configuration](/images/persistent-storage) of running a `kamu` API server along with a database,
for persistent storage of data
- New `listFlowInitiators` api to fetch all initiators of flows
- New `allPaused` method in `AccountFlowConfigs` API

Added
- New `keep_metadata_only` flag to `HardCompaction` flow. Also extended `FlowState` with `ConfigSnapshot`
and possibility to pass configuration during triggering a flow
Fixed
- Added support of `--all` flag to the `kamu delete` command
- Made recursive deletion dataset with provided `%` pattern
Changed
- `HardCompaction` configuration now is one of `Full` or `KeepMetadataOnly` variants. In case of
`KeepMetadataOnly` variant required to provide `recursive` value which will trigger downstream
dependencies compaction

Added
- Loading database components relying on CLI config
Fixed
- Panic when creating a workspace with an existing config

Fixed
- `kamu login --check` supports searching both by frontend and backend URL

Fixed
- Panic when resolving datasets in GraphQL API for unregistered accounts

Added
- Introduced MQTT protocol support (see [FetchStepMqtt](https://docs.kamu.dev/odf/reference/#fetchstepmqtt) and the new [`mqtt` example](/examples/mqtt))
- The `kamu system compact` command now accepts the `--keep-metadata-only` flag, which performs hard
compaction of dataset(root or derived) without retaining `AddData` or `ExecuteTransform` blocks
Fixed
- Panic while performing data ingesting in the derived datasets