Should everything be blazingly fast?

Devs love trends. We know that.

A trend I have seen for the last two years is to rewrite everything in Rust. Rust has some advantages: it’s fast and one could say that (thanks to its ownership and borrow checker system) it gives a good sense of safety.

It also has downsides: although some people say “it’s not harder, it’s just different” (like here and there), as a developer with 10 years of experience, I feel it’s one of the most complicated languages I have tried to learn. I think the entry level is quite high compared to many other languages and rewriting your open-source project in Rust means you probably have fewer potential contributors rather than with, say, Typescript or Python.

I was looking for a colorscheme generator for Neovim today, and found this project: colorgen-nvim. The program takes a toml file as input (a template), and generates a colorscheme for Neovim, compatible with TreeSitter.

In other words, it takes a text input and returns a text output (split in 3 or 4 files). Once.

I’m really glad and thankful this project exists, and it could be written in Python, Rust or Ruby, I don’t mind. But something in the description of the project confused me.

“Blazingly fast colorscheme generator for Neovim written in Rust”.

When did “blazingly fast” become the first quality for something that takes text as input and returns text as output, that you use once every 6 months?

I obviously understand that speed is critical for some systems, and even for some tools: I’m glad ripgrep is written in Rust and is so fast. I use it several times every day and it’s amazing to get all results almost instantaneously.

I want my Neovim to be fast.
I want my everyday tools to be fast.
I want my Terminal to be fast.
I’m always happy when my web framework gets some speed improvement.

A translator/generator I use once in a while? As soon as it generates something in less than 2 or 3 seconds, as soon as I don’t get frustrated with its slowness, I don’t mind.

I’ll be honest, you could have forgotten a sleep(0.5) function call somewhere in the program, I would have probably not seen it.

This is not to blame or criticize the great work of the LunarVim team. I’m really thankful for their tools. I’m just skeptical about this new shiny trend that doesn’t make sense to me.

If it wouldn’t make sense to add a benchmark in your README, then speed is probably not the first quality I’m looking for in your tool.