Study notes · 10% of the exam

Streams, Buffers and Backpressure

Know how Buffers share memory and encode text, how streams buffer and apply backpressure, and why `pipeline` beats chained `.pipe()` calls in production.

Key points

1
Buffer.alloc zero-fills; Buffer.allocUnsafe is faster but may contain old memory. subarray and Buffer.from(arrayBuffer) share memory, while Buffer.from(buffer | typedArray) copies.
2
String length is UTF-16 code units; Buffer.byteLength is encoded bytes. Decode chunked text with setEncoding or StringDecoder so multi-byte characters split across chunks survive.
3
write() always queues the chunk and returns false once the buffer reaches highWaterMark. Stop producing and wait for drain, or let pipeline do it for you.
4
pipeline(a, b, c, cb) forwards errors, destroys every stream on failure or premature close, and calls back once. .pipe() handles backpressure but not errors or cleanup.
5
Async iteration (for await) is pull-based and respects backpressure. Breaking out of the loop destroys the stream. Async generators can act as transforms inside pipeline.
6
A Transform must call its callback exactly once per chunk. Trailing output belongs in flush, and a missing callback stalls the whole pipeline without an error.
7
Readables start paused and keep data until someone reads it. A readable listener takes precedence over data. fetch bodies are web streams; bridge them with Readable.fromWeb.

Common traps

Buffer#slice shares memory, unlike Array and TypedArray slice, which copy.
An unhandled error event on any piped stream (for example gunzip on bad input) crashes the whole process.
Readable.from('abc') emits the whole string as one chunk; it does not iterate characters.

Read the source

Test yourself on Streams, Buffers and Backpressure

Ten questions, with the answer and explanation after each one.