---
title: "Discussion on \"What I found when I pointed a CLI recorder at a second Rust rewrite\" | Hashnode"
description: "Discussion on \"What I found when I pointed a CLI recorder at a second Rust rewrite\". I have been using behave to record how GNU coreutils 9.12 behaves, then replay the same cases against uutils, the Rust rewrite. That work has already led to a run of upstream bug reports and merged fi"
url: "https://hashnode.com/posts/what-i-found-when-i-pointed-a-cli-recorder-at-a-second-rust-rewrite/6af34335c3ae9375d6437def"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [What I found when I pointed a CLI recorder at a second Rust rewrite](https://emirmrkaljevic.hashnode.dev/what-i-found-when-i-pointed-a-cli-recorder-at-a-second-rust-rewrite?utm_source=hashnode&utm_medium=feed)

I have been using behave to record how GNU coreutils 9.12 behaves, then replay the same cases against uutils, the Rust rewrite. That work has already led to a run of upstream bug reports and merged fi

5d ago · 6 min read · emirmrkaljevic.hashnode.dev  
Read the article

-   [#rust](https://hashnode.com/tag/rust)
-   [#testing](https://hashnode.com/tag/testing)
-   [#cli](https://hashnode.com/tag/cli)
-   [#open-source](https://hashnode.com/tag/open-source)

## Written by

[Emir Mrkaljevic](https://hashnode.com/@emirmrkaljevic)

@emirmrkaljevic

  
[View profile](https://hashnode.com/@emirmrkaljevic)

## Responses 1

Markdown supported · ⌘↵ to post

[c-wind](https://hashnode.com/@c-wind)

[2d ago](https://hashnode.com/posts/what-i-found-when-i-pointed-a-cli-recorder-at-a-second-rust-rewrite/6af34335c3ae9375d6437def/comment/6af7091dcbc9a519cc5bda9f)

The `-I` default-command case is a good reminder that a rewrite can look done if you only test `xargs -I R echo R`. `printf 'hello\nworld\n' | xargs -I R` is the one that actually checks whether the replace string is a no-op when the command has no argv to start with.

The `--eof -l` pairing is the other one I'd keep in a compatibility suite: optional-argument flags are where a stricter parser silently eats the next token. Same class as `-a -foo`.

The 9.12-vs-9.4 pin is worth repeating too. Once CI is on an older GNU, a correct 9.12 case looks like a product bug, and people start skipping failures. Tagging the version-sensitive specs (`gnu-9.5`) is cleaner than weakening them.
