Mathieu Amiot

It shouldn't change for a simple reason: your User model should not change between different local strategies. The `serializeUser` / `deserializeUser` methods are here to link passport to your data...

@Ahmedrh77 `myLocalStrategy1`/`myLocalStrategy2`/`myLocalStrategy3` are placeholders for any strategy you're using, that's all. (here, you're using `Recruteur.createStrategy()`).

> What is the version of your Python and OpenCV? Which distro are you using? I guess I need to write a frame rate limiter for higher performance machines. Hi!...

Homemade parser benchmark: ``` Running target/release/deps/nitox_parser_benchmark-f72abb5a17698faf connect_parse time: [555.71 ns 561.99 ns 571.95 ns] change: [-0.1169% +0.8525% +2.1261%] (p = 0.12 > 0.05) No change in performance detected. Found 6...

Hey Kevin! I took a look at your crate and unfortunately it seems not spec compliant. A simple example is that your crate does not support `\t` as whitespace, and...

Great! I'll try when I'll have a moment, keep you updated. Did you try to check the integrity of subject names as well? Whitespace is forbidden in them, and as...

@autodidaddict I've created a benchmark suite for `nats-types` Here are the results on the same machine I ran against my home parser: ``` Running target/release/deps/nats_types_benchmark-9c16f0d712fb5429 connect_parse time: [415.51 ns 419.83...

So far I can conclude that my homemade parser is more performant on all cases except when there's JSON involved (we both use `serde_json` serialization). Really wondering why 🤔

I'd guess that we're both doing something wrong either with `nom` or `serde_json`. Something along the lines of an extra copy somewhere leading to poor performance.

@autodidaddict Soooooooo I played around with your code and made some changes, and benchmarked the difference. The changes are here: https://github.com/OtaK/nats-types/compare/feature/benchmarks...OtaK:feature/optimization Benchmark results (ran the master branch beforehand so that...