Flatten orders into lines

Toss-up

Requirement

Yesterday's four orders each hold two or three line items. Flatten them into one list of order/sku lines — every item under its order id, in source order — and print the line count. The data is in the code; both versions must print the lines shown under Expected output.

Expected output
A-101/tea-01
A-101/mug-07
A-102/pen-11
A-102/ink-02
A-102/pad-05
A-103/mug-07
A-103/lid-04
A-104/tea-01
A-104/jar-03
A-104/lid-04
10 lines from 4 orders

Side by side

RxDart

FxDart

Why they differ

One-to-many flattening is bedrock in both models, and for a synchronous payload the two spellings are the same word: Stream.expand and FxDart's flatMap both take element to iterable, splice the pieces in source order, and hand the result to a formatting map. The panels are line-for-line parallel.

The interesting divergence is just offstage. When each order's lines arrived asynchronously, the Rx side would graduate to RxDart's flatMapIterable or flatMap — inner streams, where merge order becomes a real question (interleaving by completion unless you concatenate). FxDart's async flatMap on a pulled pipeline stays in source order by construction. But that is a tier-4 story; on this in-memory job both sides express the flatten directly — the Rx panel doesn't even need an RxDart operator, core Stream carries it — and the only trace is the async main. A tie.

Benchmark

Apple M1 Max, 32 GB RAM · Dart 3.12.2 (AOT-compiled) · 2026-08-18

N = 100

Time Tie

RxDart 421 µs
FxDart 26 µs

Peak memory RxDart wins

RxDart 15.5 MB
FxDart 16.5 MB

N = 1,000,000

Time FxDart wins

RxDart 3645.4 ms
FxDart 400.1 ms

Peak memory Tie

RxDart 484.3 MB
FxDart 493.7 MB

Bars are medians of repeated timed iterations in fresh processes per side (small N is batched for timer resolution). Sides within 5% of each other — or within 0.6 ms, a difference no person can perceive — count as a tie; close relative races are re-measured up to 5 runs. In an app, anything under a few milliseconds is invisible to the user regardless of which bar is shorter. Memory is peak process RSS. The Dart VM and the dataset are identical on both sides, so the difference between the two bars is what the pipeline itself holds onto.