clap-rs VS argparse-benchmarks-rs

Seems unfair. Version 4 was almost 3 years ago and and I can't recall any issue in that time.

Major versions have good quality migration guides. If only all libraries were developed with such professionalism.

https://github.com/clap-rs/clap/blob/v4.5.0/CHANGELOG.md

Rust’s modularity and ergonomic design, combined with the powerful clap crate, allowed us to build Spin’s highly extensible Command Line Interface (CLI).

Clap: Command-Line Argument Parser for Rust

CLI parameters can be easily added to our application using the clap crate.

> To the extent there is an exception, testing-only code may be. Testing-only code has very different constraints than production code anyhow. Even then, though, I still find that refactoring problem arises, and test code needs to be refactorable too.

For me, I avoid defining anything within a function except when that thing being defined is what is being tested in a test, e.g. https://github.com/clap-rs/clap/blob/87647d268c8c27e3298b2c0...

We will be using the library for Clap - A simple-to-use, efficient, and full-featured library for parsing command line arguments and subcommands.

I recently came across this question (and associated answer) on the clap repository. The answer given is a good one. But I wanted to expand with my own findings and practices, which spurred the motivation for this post.

We can also use tuple-like struct syntax and named-field struct syntax for enum variants within our enum; this is because unlike in other OOP languages, Rust enums are actually sum types. You can read more about how powerful Rust enums are in another article we wrote here. You can have optional arguments by simply wrapping the types in Option, but if you want to add a flag to a command you can use bool, since clap recognises that flags are either there or not there. Let's have a look at what this might look like:

I began by developing a wrapper for the CurseForge API, which turned out to be a lengthy and challenging process but constituted the bulk of the work. Next, I coded the CLI, which was relatively straightforward. Instead of using the clap crate, a Rust tool for generating CLIs, I opted for the following line of code:

You can check out the argparse-benchmarks-rs though that only gives a high level summary.

So besides the git2 conversation, I'd recommend using an argument parser even if its for something basic like lexopt just for help/version, at least you'd error for unrecognized arguments and can more easily expand it in the future.

Unclear whether it has a solid advantage over other things listed at rust-cli/argparse-benchmarks-rs or whether it's just an "I didn't research what already existed" project.

I'm tempted to collect all of these benchmark repos into a github org to make them easier to find. So far I know of parser, md, argparse, and template languages.

Not strictly an answer to your question, but https://github.com/rust-cli/argparse-benchmarks-rs provides a good survey of option parsing crates in Rust.

As others have mentioned clap is one of the more popular, featureful CLI parsers though it does come at a compile time and binary size cost. We are working on improving that. See argparse-benchmarks for by-the-numbers comparison for the different parsers (ie only stats and no full feature lists). Clap is used by ripgrep, cargo, and many other tools

Compared to lexopt, which inspired this effort, clap_lex makes some different trade offs for flexibility and ergonomics which allows it to handle every case clap needs. There is room for more ergonomic improvements as the MVP was written for clap's needs. argparse-benchmarks-rs has been updated and clap_lex is roughly in line with lexopt for build-times and binary size.

See https://github.com/rust-cli/argparse-benchmarks-rs/blob/main/examples/bpaf-app/app.rs

You can use this link: https://github.com/rust-cli/argparse-benchmarks-rs

btw there are the argparse benchmarks which can be interesting points of comparison (though I haven't decided what is the criteria for being added to it yet).