Explain the difference between `pipe()` and `on('data')` when working with streams.

0-2 years experience

1. 1You need to copy the contents of one file to another using streams. Would you use pipe() or attach a 'data' listener, and why?
2. 2If you add a 'data' listener to a readable stream but never call stream.resume(), what will the program do?
3. 3What changes, if any, occur when you call pipe() on a stream that already has a 'data' listener attached?

2-5 years experience

1. 1While building a JSON‑line parser, you switched from pipe() to manual 'data' handling to log progress, but the parser now emits chunks out of order under load. What could be causing this and how would you fix it?
2. 2During a file upload you replaced pipe() with a 'data' listener to add custom headers, and the upload stalls when many users upload simultaneously. Explain why this might happen.
3. 3An error from the source stream is not reaching your error handler when you use pipe(), but it does when you listen to 'data' events. Why does this occur?

5-8 years experience

1. 1Design a streaming pipeline that validates, transforms, and writes large CSV files to a database. Discuss where you’d use pipe() versus manual 'data' handling, focusing on backpressure and error propagation.
2. 2Your service streams data from a remote API to clients over HTTP/2. How do you ensure backpressure is respected across the network boundary, and what role does pipe() play compared to 'data' events?
3. 3You need to collect per‑chunk metrics and sometimes branch the stream based on content. How would you implement this without breaking backpressure, and why might you avoid pipe() for those stages?

8+ years experience

1. 1The team is migrating a legacy codebase that heavily uses 'data' listeners to a newer architecture that prefers pipe() for composability. What architectural considerations and migration steps would you propose?
2. 2In a distributed data‑processing platform you must support both Node.js streams and external pull‑based consumers. How would you design an abstraction that can switch between pipe() and manual data handling while preserving backpressure and observability?
3. 3When evaluating a new streaming library, you need to decide whether to expose a pipe() API or an event‑emitter API to downstream teams. What long‑term maintenance, performance, and developer‑experience factors influence that decision?