orxtra v0.13.0 /services.src.orxtra.services._injection
Updated
On this page

Refresh-callback factories bridging trace readers and overseer staleness logic into the scheduler's in-memory constraint, lesson, and notepad lists, without the scheduler ever importing overseer.

#services.src.orxtra.services._injection

#services.src.orxtra.services._injection

Refresh-callback factories for scheduler injection points.

Bridges trace readers and overseer staleness logic into the scheduler's in-memory lists. Lives in the composition layer (services), which legally imports both orchestration (scheduler) and intelligence (overseer).

The scheduler receives pre-built async callbacks and never imports from overseer -- the layer boundary is maintained.

#build_constraints_refresher

python
def build_constraints_refresher(backend: StorageBackend) -> Callable[[UUID], Awaitable[list[tuple[str, str]]]]

Build a callback that reads active constraints from trace.

Returns (text, tier) tuples matching the scheduler's _active_constraints format.

#build_lessons_refresher

python
def build_lessons_refresher(backend: StorageBackend, repo_dir: Path, relevance_tags: list[str]) -> Callable[[UUID], Awaitable[list[dict[str, Any]]]]

Build a callback that reads lessons, applies staleness check.

Uses overseer's filter_stale_lessons to split fresh/stale. Returns dicts with text and stale keys, matching the scheduler's _lessons format consumed by LessonsProvider.

#build_notepad_refresher

python
def build_notepad_refresher(backend: StorageBackend) -> Callable[[UUID], Awaitable[list[NotepadEntry]]]

Build a callback that reads notepad entries from trace.

Search