-- SPDX-FileCopyrightText: 2026 Alexandra de Wit
--
-- SPDX-License-Identifier: MIT

{- | Route OpenTelemetry export failures into @katip@ through one shared throttle, so a
persistently unreachable collector is one visible warning and a periodic heartbeat rather than a
per-flush flood. The span exporter, the metric exporter, the metric batching's flush, and the SDK's
own diagnostic stream all coalesce through the same sink. None of it reaches the request path: the
SDK's batch exporter runs asynchronously, so a failed export never touches a served request. The
exporter wrappers in "Ecluse.Runtime.Telemetry" feed the sink through 'observeExportResult'.
-}
module Ecluse.Runtime.Telemetry.ExportFailure (
    ExportFailureSink,
    exportFailureSink,
    observeExportResult,
    observeFlushFault,
    installExportErrorHandler,
) where

import Ecluse.Runtime.Telemetry.ExportFailure.Internal (
    ExportFailureSink,
    exportFailureSink,
    installExportErrorHandler,
    observeExportResult,
    observeFlushFault,
 )