Rust Coreutils 0.11 Crosses 95% GNU Compatibility With Smarter Errors and Faster Builds

Developers working on Linux systems have grown used to the steady march of the uutils project. Its Rust rewrite of core command-line tools just hit version 0.11. The release brings the implementation to 95.33% compatibility with the GNU test suite. Eight more tests now pass. Failures dropped by six. Errors reached zero for the first time.

Phoronix reported the news on the same day the tag went live. The project crossed a symbolic threshold. Yet the headline improvement sits elsewhere. Error messages now look borrowed from a modern compiler.

The team integrated the Ariadne crate. It produces diagnostic output with carets pointing at the exact character causing trouble. Arguments echo back. Helpful suggestions appear. Initial support covers chmod, cut, sort and tr. More utilities will follow. Set the UUTILS_DIAG environment variable to control the behavior. The whole system can compile out for minimal builds. One stderr call handles the report. Clean. Localized. Useful for scripts and beginners alike.

But the real story runs deeper. This release caps years of work that began as an experiment and grew into a serious alternative. Canonical pushed the effort inside Ubuntu. A security audit by Zellic uncovered more than 100 issues, many involving time-of-check-to-time-of-use races. Fixes landed across 0.9 and 0.10. Ubuntu 25.10 shipped Rust versions by default for testing. The 26.04 LTS kept GNU cp, mv and rm while the remaining gaps closed. Plans pointed toward full adoption in 26.10.

Microsoft joined the picture too. At its Build conference the company announced Coreutils for Windows built on the uutils foundation. The package combines core utilities, findutils and a compatible grep into one binary. Developers can now run the same ls, cp, grep pipelines on Windows, WSL, macOS and Linux without translation. WinGet makes installation trivial. The move signals how far the Rust port has come. What started as a hobby project now ships from Redmond.

Performance work continues without fanfare. Profile-guided optimization builds are now standard for Linux, macOS and Windows releases. Some workloads see gains up to 31 percent. The release notes list concrete wins. Cp opens files once instead of twice in many cases, a 33 percent reduction in syscalls for that path. Join runs 1.37 times faster on paired files and 2.5 times faster than GNU in one benchmark. Od uses splice for skip-bytes. Tac switches to a fast_copy path. Dd aligns buffers to pages. Ptx gains four percent from a hash change. The project treats any case where it lags GNU as a bug.

Security hardening did not stop at the audit. This cycle fixed xattr TOCTOU issues in mv and cp by operating on file descriptors. Rm refuses to follow swapped symlinks during recursive removal. Chmod uses fchmodat for recursive cases and avoids following references under certain flags. Cp no longer chmods through a destination symlink. Mv avoids creating hardlinks across groups via symlinks. Stdbuf no longer leaks temporary directories. The anti-panic campaign fixed crashes in split, sort, stat, od, printf, nl, pr, csplit, dd, touch, tail, tac and who. Several deadlock and hang bugs also disappeared.

Compatibility fixes fill pages. New options appeared. Uname supports -A. Cut gained -F and –whitespace. Pr picked up –columns, -f, -J and better page handling. Numfmt, uniq, truncate, paste, join, id, unexpand, expand, mkdir, more, fold, sort, stat, od, date and others saw behavior aligned with GNU. Error messages now include filenames and real OS errors instead of vague text. Documentation checks ensure option lists match. The test suite grew by two cases. The pass rate still climbed.

Cross-platform support broadened again. Windows gained kill and uptime plus several test operators. WASI added touch, cp symlink handling and expr. Df falls back to statfs. Uptime works without procfs. Stty and chgrp stay gated where needed. The dependency list shrank. Byteorder, same-file, filetime and others gave way to std, rustix and windows-sys. Internationalization moved forward with translatable messages, locale-aware ls sorting, collation in comm, multibyte support in nl and cut.

Contributors poured in. Three hundred fifty-eight pull requests arrived from twenty-six first-time participants. The project now counts more than one thousand total contributors. August 2026 set records with 476 commits from fifty-seven authors. Lead developer Sylvestre Ledru has presented the work at FOSDEM multiple times. His talks highlight how early critics underestimated the effort. Many predicted Rust could never match C performance or GNU behavior. The numbers keep proving them wrong. Ubuntu adoption provided real-world data. Bugs surfaced and vanished quickly.

The project no longer stops at coreutils. Plans exist to rewrite diffutils, findutils, util-linux pieces and more under the uutils umbrella. The same drop-in replacement strategy applies. Maintainers invite testing through an online playground. Translation help is welcome on Weblate. GitHub sponsors can support the work directly.

So the Rust version stands at a crossroads. It offers nicer errors, measurable speed gains, stronger security and near-parity with decades of GNU development. Distributions can choose. Some already have. Others watch the final gaps close. For systems programmers and package maintainers the message is clear. A viable, memory-safe alternative exists. And it keeps getting better with every release.

Those curious can read the full release notes on GitHub. The error diagnostics explanation lives on the uutils blog. Earlier coverage of the security audit and Ubuntu plans appears in Phoronix articles from April 2026. Microsoft’s Windows port drew attention in Bleeping Computer and additional Phoronix reporting. The momentum shows no sign of slowing.