Skip to content

Error Handling with TRY/CATCH

Use TRY / CATCH to handle failures and optionally retry operations.

Catching failures

TRY riskyCall()
CATCH(e) -> { err: e.message };

If riskyCall() throws, the fallback block produces an error object.

Retrying

Add RETRY n TIMES to re-run the expression before giving up.

TRY fetch(input.id)
CATCH(e) RETRY 3 TIMES -> { err: "failed" };

The call is attempted three additional times. If all attempts fail, the fallback runs.

With backoff

TRY CALL remoteService(input) -> response
CATCH(err) RETRY 2 TIMES BACKOFF 500ms -> {
    error: err.message ?? "service failed"
};
Adds a fixed backoff between attempts. (Host must support CALL and backoff semantics.)

Guardrails with ASSERT

Combine ASSERT to fail fast on bad states: 

LET result = TRY risky(input) CATCH(e) -> NULL;
ASSERT(result != NULL, "Risky call failed");

Tips

  • TRY wraps expressions; use it sparingly and close to the failing call.
  • Provide clear fallback values/objects for downstream logic.
  • BACKOFF units follow host runtime defaults (e.g., ms).
  • Pair with CHECKPOINT when you need progress markers around retries.

Try it

Open the pipeline-health-gating example in the playground to see ASSERT and tracing in action; adapt it to wrap TRY/CATCH around external calls in your own snippets.