Module retry 

Retrying a failed operation. Not available on wasm32 — tokio’s time driver, which the backoff needs, does not run there. Retrying an operation that failed for a reason worth retrying.

The crate has carried a complete retryability classification with no mechanism: DataflowError::retryable sorts every variant, and ErrorInfo has retry_attempted / retry_count fields, but no engine code path acts on any of it. Every host has written the same loop.

This module supplies the loop. It is deliberately not engine-level automatic retry: the engine cannot know which handlers are idempotent — an SMTP send that times out after DATA is indistinguishable from one that succeeded, and retrying duplicates the mail. Whether to retry stays a per-handler, per-call-site decision; the crate just supplies the correct loop for those that opt in.

Not available on wasm32

Backoff needs a timer, and tokio’s time driver does not run on wasm32-unknown-unknown. The whole module is cfg-gated off that target.

Structs

RetryPolicy

How hard to retry, and for how long overall.

Functions

retry_with_attempts

As retry_with_policy, also reporting how many attempts were made.

retry_with_policy

Run operation, retrying while it fails retryably and budget remains.