Page 3 of 8~104 min topic

Streaming responses

Build the first working SSE token relay

Page 3 implements the shortest complete path for the SSE relay for model tokens with inspectable intermediate values.

~13 min this pageImplementation

1Learn the idea

Read

Implement the minimal working path

Build only what the claim requires: client receives ordered data events and a terminal done event; cancel stops further writes. Prefer boring, deterministic code over frameworks you cannot yet explain. Run the path twice; identical output on this fixture is a feature, not a lack of creativity.

Read

Run the working path

export async function run(input: unknown, deps: Dependencies) {
  const started = performance.now();
  const validated = deps.parseInput(input);
  const raw = await deps.execute(validated); // read the provider response body with a Web ReadableStream and emit framed SSE events
  const value = deps.parseOutput(raw);
  return { value, durationMs: Math.round(performance.now() - started) };
}

Expected evidence: The client renders “Hello 👋” in order, receives done once, and closes cleanly.. Read each printed intermediate as part of the argument that the path works—not as decoration.

Read

Trace one input end to end

Narrate the journey from raw input to result for a single example from fake chunk iterator yielding three tokens then done. If you cannot name an intermediate, the implementation is still too opaque for this lab. Only after this path is solid should you generalize data sources or UI.

Read

Lab notebook: intermediates worth printing

While implementing the SSE token relay, print or log at least three intermediates that map to the claim (client receives ordered data events and a terminal done event; cancel stops further writes). Good intermediates are values a teammate could recompute with a calculator or diff. Bad intermediates are framework traces you cannot explain.

Re-run with fake chunk iterator yielding three tokens then done twice. If the second run differs, either the path is nondeterministic (document the seed) or you have hidden global state—both are lab bugs until named.

Read

Worked judgment

Stop adding features once the path supports stream tokens to a client while honoring cancel and partial failure. Extra UI, extra tools, or extra models belong in later chapters. The mastery bar for this page is simply: a deterministic end-to-end path with intermediates.

Read

Why this stage matters for the SSE token relay

At the implementation stage for streaming-responses, the job is narrower than finishing a product demo. You are creating one progressive evidence piece about fake chunk iterator yielding three tokens then done that later pages inherit without redefining success. Keep that fixture small enough to inspect by hand, keep outputs copy-pasteable as text, and refuse to narrate this baseline as if it were a production SLA: non-streaming JSON endpoint latency for the same prompt.

For this page specifically, success looks like a deterministic path with printed intermediates while still centering the user decision to stream tokens to a client while honoring cancel and partial failure. If you cannot point to a file, command, or assertion that proves that for the SSE token relay, stay on this page instead of advancing.

Glossary: tool · Glossary: structured output · Cheatsheet: production ops signals

Previous · Next

Go deeper

Before you start

Why this matters

Without running code, predict the final output for fixture fake chunk iterator yielding three tokens then done. Name one intermediate value that would prove the prediction. Then answer: what could look successful while actually being wrong at this stage for the SSE token relay?

In the wild

See how this idea shows up as a product and a company — then come back to the lesson. Skills transfer across vendors.

Check your understanding

Page assessment

Answer from memory. Completion is saved from this evidence, not from opening the next page.

1. Can you narrate every intermediate value?
2. Is the fixture deterministic and independently inspectable?
3. Did you avoid framework behavior you cannot explain yet?
4. Does the output still support the decision: stream tokens to a client while honoring cancel and partial failure?

All responses are required.