structopt VS anyhow

Hope you are aware that structopt is in maintenance mode and is merged into clap as of v3.

As I understood you need to implement a command line argument parser for that you can use clap https://github.com/clap-rs/clap or structopt https://github.com/TeXitoi/structopt.

I feel like discovering moves like this is a weakness in the ecosystem today. You can check out some of our discussion on raising visibility

Something I've been giving thought to is how to help structopt users discover that clap3 is their upgrade path. We've put notices in the structopt repo but cargo upgrade and docs.rs won't say anything. See https://github.com/TeXitoi/structopt/issues/525 for more ideas we're considering.

For myself, I have found serde.rs really useful for undertanding their derives while I've always been frustrated with finding anything in structopt's documentation, so I modeled it more off of serde. This ended up both being in structure and not being in docs.rs. I think it really was the structure that was the frustration point for me but there was interest elsewhere in moving stuff out of docs.rs.

I think it would be nice to have a comparison to clap-derive and/or structopt in the README, as that is what I expect most users would compare this to. The subcommand handling looks especially cumbersome compared to deriving on structs and enums.

However, the one place I'm a little curious to rewrite things is the CLI... every time we have to deal with cobra I long for Rust's structopt.

Have you tried https://github.com/TeXitoi/structopt ?

It is written in Rust using the structopt and swayipc crates. It is published on crates.io. The repository is hosted on GitLab. The repository is mirrored on GitHub.

I use structopt, which itself uses clap.

Depending on your use case, thiserror and/or anyhow.

There is ? as well as the anyhow(https://github.com/dtolnay/anyhow) crate that deals with long nested result chains.

There are crates out there that help you reduce this boiler plate. thiserror is good for creating custom errors and color-eyre or anyhow are good for dynamic errors.

Give anyhow a look.

I've brought anyhow::Result into scope, making error handling super easy to use. We don't need to specify all our Error types. It can automatically convert any errors that implement std::error::Error, which should be all of them. If an error propagates all the way up to main(), we'll get all the info it's captured printed to stdout.

Alternatively, if you've got a lot of error types and you're outside a library (so directly in a binary where you don't plan to reuse code elsewhere) you can use anyhow. This gives you an error type you can basically propagate any other error through. On top of that you can attach context information at every return. It's basically a more complicated Result>.

One approach is to define a separate error type for each crate and then use anyhow, eyre or Box to wrap the error, whever a function can return errors originating in several different crates.

Have you tried anyhow and thiserror for making it as simple as .context("Message") or ? to type-convert and propagate errors up the call stack?