Skip to content

consider removing custom partition_point #130

@emberian

Description

@emberian

It would be dandy to get rid of the only usage of unsafe :)

I ran a simple experiment where I ran cargo bench --bench tmi, then replaced the partition_point implementation with a simple slice.partition_point(pred), and got decent results, nothing like the original regressions:

rustc 1.87.0-nightly (249cb8431 2025-03-12)

Image

The impact seems small, but I don't know the importance of this primitive on the system performance.

Metadata

Metadata

Assignees

Type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions