Feed aggregator

Article URL: https://bsky.app/profile/leftistwonk.bsky.social/post/3mnbd6txlus2c

Comments URL: https://news.ycombinator.com/item?id=48366144

Points: 1

# Comments: 0

Article URL: https://www.theverge.com/column/939026/user-replaceable-batteries-eu-european-union-legislation

Comments URL: https://news.ycombinator.com/item?id=48366133

Points: 2

# Comments: 0

Article URL: https://which-llm-model.netlify.app/

Comments URL: https://news.ycombinator.com/item?id=48366110

Points: 1

# Comments: 0

Article URL: https://stackoverflow.blog/2026/05/21/coding-agents-are-giving-everyone-decision-fatigue/

Comments URL: https://news.ycombinator.com/item?id=48366108

Points: 1

# Comments: 0

Article URL: https://worksinprogress.co/issue/samurai-city/

Comments URL: https://news.ycombinator.com/item?id=48366090

Points: 1

# Comments: 0

Article URL: https://pypi.org/project/shadowprotect/0.1.0/

Comments URL: https://news.ycombinator.com/item?id=48366086

Points: 2

# Comments: 0

Article URL: https://blog.jetbrains.com/ai/2026/06/mellum2-goes-open-source-a-fast-model-for-ai-workflows/

Comments URL: https://news.ycombinator.com/item?id=48366057

Points: 3

# Comments: 0

Article URL: https://discourse.julialang.org/t/ann-http-jl-2-0-release-and-new-package-reseau-jl/137323

Comments URL: https://news.ycombinator.com/item?id=48366022

Points: 2

# Comments: 0

Article URL: https://hackaday.com/2026/05/21/magnets-are-bad-for-hardware-again/

Comments URL: https://news.ycombinator.com/item?id=48366003

Points: 2

# Comments: 0

Article URL: https://research.nvidia.com/labs/amri/projects/video-fdb/

Comments URL: https://news.ycombinator.com/item?id=48365984

Points: 2

# Comments: 0

Article URL: https://variety.com/2026/biz/tech/florida-sues-openai-sam-altman-1236764066/

Comments URL: https://news.ycombinator.com/item?id=48365951

Points: 4

# Comments: 1

Article URL: https://david.alvarezrosa.com/posts/self-hosting-on-the-dark-web/

Comments URL: https://news.ycombinator.com/item?id=48365945

Points: 2

# Comments: 0

Article URL: https://thezvi.substack.com/p/claude-opus-48-is-honestly-better

Comments URL: https://news.ycombinator.com/item?id=48365939

Points: 1

# Comments: 0

Article URL: https://startingstrength.com/article/squats-presses-and-deadlifts-why-gyms-dont-teach-the-only-exercises-you-need

Comments URL: https://news.ycombinator.com/item?id=48365909

Points: 2

# Comments: 0

Article URL: https://geerttheys.substack.com/p/llms-vs-human-engineers-who-really

Comments URL: https://news.ycombinator.com/item?id=48365905

Points: 3

# Comments: 0

Article URL: https://blog.janestreet.com/strace-ui-bonsai-term-and-the-tui-renaissance/

Comments URL: https://news.ycombinator.com/item?id=48365904

Points: 2

# Comments: 0

Here are the answers for The New York Times Mini Crossword for June 2.

I first came across the paper “Optimal Bounds for Open Addressing Without Reordering” through a Quanta Magazine YouTube video a few months back. I went looking for an official implementation and couldn’t find one, so I decided to try implementing the paper's Elastic Hashing and Funnel Hashing in Rust.

To that end, I build opthash, a Rust library providing ElasticHashMap and FunnelHashMap implementations (and more recently HashSet variants). They are at API parity with std::collections::HashMap and HashSet.

Initially, the data layout was a relatively straightforward implementation of the paper, with levels and buckets represented as nested structs that each managed their own allocation. Later, I moved to a flatter arena layout with one allocation for the backing control/data regions, and smaller descriptor structs holding pointers and metadata for each level. That ended up being noticeably faster. The intuition I went for is to arrange control bytes contiguously to maximize cache locality, since majority of instructions for probing was done on the control bytes.

Most the low-level details is inspired by hashbrown/SwissTable, such as 7-bit control bytes, SIMD scans over groups of control bytes, power-of-two sizing, test cases, and foldhash as the default hasher. hashbrown is included in the benchmarks as the performance ceiling, and std::HashMap is the baseline.

I also added Python bindings as a learning exercise. I aimed for parity with Python dict and set, but I quickly realized crossing through PyO3 adds more overhead than expected...

I would gladly appreciate feedback, especially about hash table construction, PyO3 bindings, or benchmarking methodology. I have learned a lot about Rust's language features (and crossing into unsafe territory) from this project, and I'm sure there are still many things to improve.

Comments URL: https://news.ycombinator.com/item?id=48365720

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48365717

Points: 1

# Comments: 1