BackendOutcome#

class scikitplot.corpus.BackendOutcome(status, value, backend, attempted, missed, errors, skipped=(), policy='resilient', selected_capabilities=(), active_capability=None, skip_details=(), requested_backend=None, allow_fallback=True)[source]#

Structured, serialisable outcome from run_backend_chain.

Notes

User. The fallback vocabulary (R2-5) shared with the vector backends and SemanticChunker: requested_backend is the first candidate the policy allows to run (a candidate the policy skips – network disabled, not ready – was never eligible, like an absent backend under "auto"; skipped and skip_details record it), backend the one that produced the value, allow_fallback the policy’s switch (False: no other candidate is attempted), and fallback_reason why they differ: "failed" (it raised) or "no_result" (it returned nothing usable).

Parameters:
active_capability: str | None = None#
allow_fallback: bool = True#
attempted: tuple[str, ...][source]#
backend: str | None[source]#
capability_view(*, registry=None)[source]#

Return readiness records with selected/active runtime provenance.

Parameters:

registry (CapabilityRegistry | None)

Return type:

dict[str, dict[str, Any]]

errors: tuple[ErrorRecord, ...][source]#
property fallback_reason: str | None#

Why backend is not requested_backend, or None.

Returns:
str or None

"failed" or "no_result" for the requested candidate’s fate; None when it produced the value, or when no candidate was eligible.

missed: tuple[str, ...][source]#
policy: str = 'resilient'#
requested_backend: str | None = None#
selected_capabilities: tuple[str, ...] = ()#
skip_details: tuple[BackendSkip, ...] = ()#
skipped: tuple[str, ...] = ()#
status: BackendStatus[source]#
property succeeded: bool#

Whether some backend completed successfully.

to_dict()[source]#

Return a JSON-compatible representation without live exceptions.

Return type:

dict[str, Any]

value: _T[source]#