What's this about?
The Polars 2.0 release candidate makes the streaming engine the default for every LazyFrame query — expected to be roughly 5× faster — and stops guaranteeing row order on joins and group-bys unless you ask for it.
What exactly is Polars?
The Rust-powered dataframe library that has been steadily winning converts from pandas on speed and API discipline. A "streaming engine" processes data in chunks rather than loading everything into memory — bigger-than-RAM datasets, better parallelism.
What's the community saying?
HN 384 points, 129 comments
- Genuine affection for a boring major release: "we hope it to be a boring experience for you" quoted approvingly — 2.0 exists to fix old design decisions, not add features.
- The one sharp objection: unordered-by-default is a documented source of bugs in *scientific* pipelines, where non-determinism silently corrupts reproducibility.
- Testimonials on why teams switched: pandas pushes problems to runtime with hidden heuristics; Polars fails loudly at plan time.
- A sidebar on the ecosystem shifting: DuckDB's DuckLab was just acquired by AWS, and people are re-evaluating the whole dataframe/SQL landscape.