It might be just the write up, but the article reads like he was playing with the benchmarks until he got a result where Java was faster. I don’t think anythin

It might be just the write up, but the article reads like he was playing with the benchmarks until he got a result where Java was faster. I don’t think anything malicious has happened, just pointing out that benchmarking is hard, and changing parameters just to see what happens is not benchmarking. I haven’t gone deep enough into the code and the problem, but it looks like this: What if we swap the Rust HashMap implementation, as they say the standard one I was using applies a cryptographically-safe hash that the JVM doesn’t Is quite a distinction, especially for the latter cases with lengthier phone numbers.