- Enhanced deployment request preparation to include logs configuration.
- Introduced logs writeback configuration management in the database layer.
- Added new queries for upserting and deleting logs writeback configurations.
- Implemented logs writeback task to write query logs to Snowflake.
- Updated data source adapters to support logs table creation and record insertion.
- Integrated logs writeback task trigger in message post-processing workflow.
- Defined schemas for logs configuration and writeback records in shared packages.
This commit lays the groundwork for improved logging capabilities, allowing organizations to track query logs effectively.
- Add concurrent embedding generation (5 batches concurrently)
- Add concurrent Turbopuffer upserts (3 batches concurrently)
- Fix dynamic SQL identifier quoting for different databases (Snowflake, PostgreSQL, MySQL, etc.)
- Remove unnecessary ORDER BY clause when joining with cached datasets
These optimizations significantly reduce processing time for large datasets by processing
multiple batches in parallel instead of sequentially.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Implemented tests for Slack notification skip logic based on issue and assumption conditions.
- Updated the main task to skip notifications when no issues are found and no major assumptions exist.
- Enhanced workflow to handle undefined `isSlackFollowUp` correctly, ensuring proper message formatting and notification sending.
- Added new deployment functionality to handle both models and documentation in a single request.
- Introduced new schemas for unified deployment requests and responses using Zod.
- Updated deployment handlers to process and validate models and docs together.
- Enhanced logging to provide detailed information about deleted models and docs during deployment.
- Refactored existing code to integrate the new deployment structure, ensuring backward compatibility with previous functionalities.