What makes your code slow? How do you make it faster? And how do you prove it?
This talk will describe my adventures benchmarking and optimizing ordered hashes in Perl, culminating in the release of Hash::Ordered — which outperforms all other CPAN alternatives, often by a substantial margin. We will cover:
- How to customize Benchmark.pm
- How and why to benchmark at different scales
- Why tied anything in Perl is a horrible idea
- How ordered hashes got faster from a simple algorithm change