Polars 2.0 Release Candidate Introduces Streaming Engine as Default

Polars 2.0 Release Candidate Announced

Polars has announced the release candidate for version 2.0, with the final release expected in the coming weeks. The version bump addresses past design decisions and introduces new defaults to benefit a broader user base.

Streaming Engine Becomes Default

The most significant change in Polars 2.0 is the default activation of the streaming engine for all LazyFrame queries. This modification is projected to lead to substantial improvements in memory efficiency and query performance, potentially making the streaming engine up to five times faster in aggregate.

This change required a major version increment because the streaming engine does not guarantee row order by default for certain operations like join, group_by, and unpivot. Users who require observable row order can enable it by setting `maintain_order=True` for specific operations.

Configuration Options for Engine Affinity

Users can still opt to use the older "in-memory" engine as their default by configuring engine affinity globally or on a per-query basis. This provides flexibility for those who prefer the previous behavior or need to ensure row order without explicit flags.

Emphasis on Stricter Behavior

Polars 2.0 continues the project's focus on strictness and failing fast. The goal is to raise errors early in the development process rather than allowing implicit behaviors on data mismatches to hide potential bugs, a principle that has become more important with the rise of AI-driven development.