- Added tests for AI service to ensure it operates independently of knowledge configuration, including scenarios with missing or corrupt embedding models.
- Updated existing tests to prepare embedding models before execution to ensure consistent test environments.
- Introduced new utility functions for managing embedding models in test setups.
- Improved error handling in knowledge retrieval tests to verify behavior when embedding models are unavailable.
- Refactored test harness to streamline embedding model preparation and ensure proper initialization before running tests.
Keep Mac delta assertions strict, and only use ready-state after-count fallback on Windows when both measured deltas stay at 0 for a warm embedding worker.
Windows OS thread counts after creating an ORT session are intraOp-1 because
the calling thread is part of the pool. Foundation and settings E2E asserts
now use that floor while still requiring the slow/fast delta gap to match
the intraOp gap.
E2E no longer depends on Win32 CommandLine containing ncurator-emb-N.
EmbeddingRuntimeClient records the utility process PID after spawn, and
getEmbeddingStatus returns that PID plus intraOp for process discovery.
Co-authored-by: Yoan <Yoan98@users.noreply.github.com>
- Replace `runtimeSettingsOverride` with `streamingModeOverride` in multiple test files to streamline the configuration of LLM streaming modes.
- Update references to `aiRuntimeSettings` to use the new `appSettings` structure in relevant tests.
- Introduce new tests for workspace preparation and indexing to ensure sequential execution and proper handling of workspace jobs.
- Adjust SQLite migration tests to reflect changes in schema versioning and settings persistence.
- Modify system notification service tests to align with the updated app settings structure.
- Implement tests for KB indexing blocks search to handle errors during indexing.
- Add tests for embedding error recognition and initialization message mapping.
- Create tests for embedding runtime contract to ensure proper request serialization.
- Introduce tests for embedding service initialization and error handling.
- Develop tests for model download source contract to validate metadata and file handling.
- Add comprehensive tests for vector download service to ensure single-task execution and proper progress reporting.
- Implement workspace knowledge cache tests to verify content isolation and statistics tracking.
hy