Spice v0.18.1-beta (Sep 23, 2024)
Announcing the release of Spice v0.18.1-beta. 🏎️
The v0.18.1-beta release continues to improve runtime performance and reliability. Performance for accelerated queries joining multiple datasets has been significantly improved with join push-down support. GraphQL, MySQL, and SharePoint data connectors have better reliability and error handling, and a new Microsoft SQL Server data connector has been introduced. Task History now has fine-grained configuration, including the ability to disable the feature entirely. A new spice search CLI command has been added, enabling development-time embeddings-based searches across datasets.
Highlights in v0.18.1-beta
Join push-down for accelerations: Queries to the same accelerator will now push-down joins, significantly improving acceleration performance for queries joining multiple tables.
Microsoft SQL Server Data Connector: Use from: mssql: to access and accelerate Microsoft SQL Server datasets.
Example spicepod.yml:
datasets:
- from: mssql:path.to.my_dataset
name: my_dataset
params:
mssql_connection_string: ${secrets:mssql_connection_string}
See the Microsoft SQL Server Data Connector documentation.
Task History: Task History can be configured in the spicepod.yml, including the ability to include, or truncate outputs such as the results of a SQL query.
Example spicepod.yml:
runtime:
task_history:
enabled: true
captured_output: truncated
retention_period: 8h
retention_check_interval: 15m
See the Task History Spicepod reference for more information on possible values and behaviors.
Search CLI Command Use the spice search CLI command to perform embeddings-based searches across search configure datasets. Note: Search requires the ai feature to be installed.
Refresh on File Changes: File Data Connector data refreshes can be configured to be triggered when the source file is modified through a file system watcher. Enable the watcher by adding file_watcher: enabled to the acceleration parameters.
Example spicepod.yml:
datasets:
- from: file://path/to/my_file.csv
name: my_file
acceleration:
enabled: true
refresh_mode: full
params:
file_watcher: enabled
Breaking Changes
The Query History table runtime.query_history has been deprecated and removed in favor of the Task History table runtime.task_history. The Task History table tracks tasks across all features such as SQL query, vector search, and AI completion in a unified table.
See the Task History documentation.
Dependencies
- DataFusion: Upgraded from v41 to v42.
- Apache Arrow: Upgraded from v52 to v53.
- DuckDB: Upgraded from v1.0 to v1.1.
Contributors
- @phillipleblanc
- @Jeadie
- @lukekim
- @sgrebnov
- @peasee
- @Sevenannn
- @ewgenius
- @slyons
New Contributors
- @slyons made their first contribution in https://github.com/spiceai/spiceai/pull/2724
What's Changed
- Update Helm Chart for 0.18.0-beta release by @sgrebnov in https://github.com/spiceai/spiceai/pull/2711
- Use a single instance for all DuckDB accelerated datasets by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2669
- Dependabot upgrades by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2715
- Use a single instance for all SQLite accelerated datasets by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2720
- Prepare for v0.18.1-beta release by @sgrebnov in https://github.com/spiceai/spiceai/pull/2692
- For GraphQL, remove necessity of `json_pointer` and improve error messaging. by @Jeadie in https://github.com/spiceai/spiceai/pull/2713
- Postgres accelerator benchmark test by @Sevenannn in https://github.com/spiceai/spiceai/pull/2652
- Trace query result while running benchmark tests by @sgrebnov in https://github.com/spiceai/spiceai/pull/2684
- Early check EmbeddingConnector if embedding models do not exist by @Jeadie in https://github.com/spiceai/spiceai/pull/2717
- Move table creation for spice_sys_dataset_checkpoint to DatasetCheckpoint::try_new by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2732
- Don't load tools immediately by @Jeadie in https://github.com/spiceai/spiceai/pull/2731
- Renable accelerator federation on trunk by @Sevenannn in https://github.com/spiceai/spiceai/pull/2725
- Fixing Data Connectors link in README.md by @slyons in https://github.com/spiceai/spiceai/pull/2724
- Enable rehydration tests for DuckDB by @sgrebnov in https://github.com/spiceai/spiceai/pull/2729
- Check pageInfo is correct at initialisation of GraphQL connector by @Jeadie in https://github.com/spiceai/spiceai/pull/2730
- Microsoft SQL Server data connector initial support by @sgrebnov in https://github.com/spiceai/spiceai/pull/2741
- Add `spice search` CLI command by @lukekim in https://github.com/spiceai/spiceai/pull/2739
- Update threat model by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2738
- Upgrade to Arrow 53, DataFusion 42 and DuckDB 1.1 by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2744
- Update datafusion table provider patch by @Sevenannn in https://github.com/spiceai/spiceai/pull/2747
- feat: Add enabled config option for task_history by @peasee in https://github.com/spiceai/spiceai/pull/2758
- Remove v0.18.0-beta from the Roadmap by @sgrebnov in https://github.com/spiceai/spiceai/pull/2748
- Fix spark-connect to use native roots for TLS again by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2766
- Fix benchmark test - Install default crypto provider by @Sevenannn in https://github.com/spiceai/spiceai/pull/2752
- Resolve primary keys for datasets with catalog or schema by @Jeadie in https://github.com/spiceai/spiceai/pull/2749
- MSSQL: include table name in schema retrieval error by @sgrebnov in https://github.com/spiceai/spiceai/pull/2746
- File Format parsing for Document tables, support for docx + pdf by @Jeadie in https://github.com/spiceai/spiceai/pull/2740
- Add Document parsing to Sharepoint connector. by @Jeadie in https://github.com/spiceai/spiceai/pull/2760
- Execution plan with BinaryExpr predicates pushdown support for MS SQL by @sgrebnov in https://github.com/spiceai/spiceai/pull/2768
- Update datafusion patch by @Sevenannn in https://github.com/spiceai/spiceai/pull/2772
- Support for standalone config parameters for MS SQL by @sgrebnov in https://github.com/spiceai/spiceai/pull/2773
- Utilize DataConnectorError for MySQL Data Connector Errors by @Sevenannn in https://github.com/spiceai/spiceai/pull/2759
- Add Score to search results by @lukekim in https://github.com/spiceai/spiceai/pull/2774
- Don't call GetComponentStatuses when --metrics not enabled by @Jeadie in https://github.com/spiceai/spiceai/pull/2779
- Implement better error handling for spicepods by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2767
- Make integration tests more robust by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2782
- Query results streaming support for MS SQL by @sgrebnov in https://github.com/spiceai/spiceai/pull/2781
- Update benchmark snapshots by @Sevenannn in https://github.com/spiceai/spiceai/pull/2778
- For Sharepoint connector, if client_secret and auth_code are both provided, default to auth_code by @Jeadie in https://github.com/spiceai/spiceai/pull/2780
- Add modified pk/indexes scenario to rehydration tests by @sgrebnov in https://github.com/spiceai/spiceai/pull/2743
- Run benchmarks on Wed, Fri, Sat, and Sun. by @lukekim in https://github.com/spiceai/spiceai/pull/2786
- Update PULL_REQUEST_TEMPLATE.md to include a section for Documentation by @slyons in https://github.com/spiceai/spiceai/pull/2785
- Add E2E test for MS SQL data connector by @sgrebnov in https://github.com/spiceai/spiceai/pull/2788
- More types support for MS SQL data connector by @sgrebnov in https://github.com/spiceai/spiceai/pull/2789
- feat: Add captured_output option for task_history by @peasee in https://github.com/spiceai/spiceai/pull/2783
- Add ability to refresh when file data connector detects changes by @phillipleblanc in https://github.com/spiceai/spiceai/pull/2787
- Propagate MySQL invalid table name error by @Sevenannn in https://github.com/spiceai/spiceai/pull/2776
- feat: Add retention options for task_history config by @peasee in https://github.com/spiceai/spiceai/pull/2784
- fix: Move task history check after query history creation by @peasee in https://github.com/spiceai/spiceai/pull/2793
- MS SQL connector should ignore all unsupported types by @sgrebnov in https://github.com/spiceai/spiceai/pull/2795
- Improve Sharepoint DX by @Jeadie in https://github.com/spiceai/spiceai/pull/2791
- Replace query history with task history by @peasee in https://github.com/spiceai/spiceai/pull/2792
- Fix datasets_health_monitor spice.runtime.task_history not found warning by @sgrebnov in https://github.com/spiceai/spiceai/pull/2805
- Upgrade macOS x86_64 test runner to macOS 13.6.9 Ventura by @sgrebnov in https://github.com/spiceai/spiceai/pull/2803
- Update acknowledgements by @github-actions in https://github.com/spiceai/spiceai/pull/2808
- Add mssql to the list of supported data connectors by @sgrebnov in https://github.com/spiceai/spiceai/pull/28
**Full Changelog**: https://github.com/spiceai/spiceai/compare/v0.18.0-beta...v0.18.1-beta
Resources
Community
Spice.ai started with the vision to make AI easy for developers. We are building Spice.ai in the open and with the community. Reach out on Slack or by email to get involved.
- Twitter: @spice_ai
- Slack: spiceai.org/slack
- Telegram: Spice AI Discussion
- Reddit: https://www.reddit.com/r/spiceai
- Email: hey@spice.ai
