tirth8205/code-review-graph · warning · RuntimeError

post-processing reported warnings: {details}

Error message

post-processing reported warnings: {details}

What it means

Counterpart to the errors check: _raise_watch_postprocess_warnings promotes non-empty 'warnings' from a watch update's post-processing stage into a RuntimeError, treating degraded results as failures.

Source

Thrown at code_review_graph/incremental.py:1590

    """Fail the watch boundary when an incremental update reports errors."""
    errors = result.get("errors") or []
    if not errors:
        return
    details = "; ".join(
        f"{error.get('file', 'unknown')}: {error.get('error', 'unknown error')}"
        for error in errors
    )
    raise RuntimeError(f"{context} reported errors: {details}")


def _raise_watch_postprocess_warnings(result: object) -> None:
    """Treat structured post-processing warnings as a failed watch update."""
    if not isinstance(result, dict):
        return
    warnings = result.get("warnings") or []
    if warnings:
        details = "; ".join(str(warning) for warning in warnings)
        raise RuntimeError(f"post-processing reported warnings: {details}")


# ---------------------------------------------------------------------------
# Watch scheduling and supervision
# ---------------------------------------------------------------------------

# A single recursive watch on the repository root makes the OS register one
# watch per directory in the tree — including every temp directory a build tool
# churns through inside ``target/`` or ``node_modules/``.  Planning the watches
# ourselves keeps ignored trees off the OS watch list entirely.  See: #811.
_WATCH_PLAN_DEPTH = int(os.environ.get("CRG_WATCH_PLAN_DEPTH", "3"))
_MAX_WATCH_SCHEDULES = int(os.environ.get("CRG_MAX_WATCH_SCHEDULES", "24"))
# Splitting a watch costs one watchdog emitter, so it has to buy more than it
# costs: an ignored tree is only worth excluding once it holds this many
# directories.  A lone ``__pycache__`` is not worth a thread; ``target/`` is.
_WATCH_SPLIT_MIN_DIRS = int(os.environ.get("CRG_WATCH_SPLIT_MIN_DIRS", "4"))
_WATCH_HEALTH_INTERVAL = float(os.environ.get("CRG_WATCH_HEALTH_INTERVAL", "10"))
_WATCH_STOP_TIMEOUT = 10.0

View on GitHub (pinned to b58668751a)

Solutions

1. Read the warning details in the message to see which postprocess step flagged issues
2. Trigger a full rebuild of the graph to reset state
3. If warnings recur on the same input, report with the repo snapshot that triggers it

null

def has_postprocess_warnings(result: dict) -> bool:
    return isinstance(result, dict) and bool(result.get("warnings"))

try:
    result = watch(...)
except RuntimeError as e:
    if "post-processing reported warnings" in str(e):
        log.warning("watch postprocess degraded: %s", e)
        trigger_full_rebuild()
    else:
        raise

When it happens

Trigger: A watch/process update whose result contains entries under 'warnings' — post-processing (e.g. relationship resolution, edge consolidation) flagged anomalies it could not fully resolve.

Common situations: Corrupted or partially-updated graph state after interrupted updates, inconsistent edges after rapid file churn, or bugs in post-processing logic surfacing on unusual graph shapes.

Related errors

• {context} reported errors: {details}

AI-assisted analysis of tirth8205/code-review-graph@b58668751a (2026-08-28). Data as JSON: /api/errors/5576f3887219951c. Report an issue: GitHub.