Post by @w8emv@a2mi.social

2026-05-04 21:11 UTC

@hyc @lcamtuf One thing that the Rust rewrite of coreutils tried to do was to prove that it was making steady progress by the number of test cases originating from GNU coreutils that it could pass. I very much suspect that there's a whole host of race condition tests that made it into the test corpus late in the game. Test-driven rewrite has its limits. Note the uptick in failures at the very right edge of the graph, they are currently under 90% tests successful. #coreutils #uutils

Replies (3)

@hyc@mastodon.social 2026-05-04 21:24

@w8emv @lcamtuf a rather damning graph for Canonical, choosing to deploy when they knew it was less than 90% compatible. Even if they were passing 100% of the test suite you'd expect problems with edge cases that hadn't been covered yet.

Open ##1979696
@bob_zim@infosec.exchange 2026-05-04 22:18

@w8emv @hyc @lcamtuf Is that ~700 tests for all of coreutils? That seems alarmingly low.

Open ##1981377
@theeclecticdyslexic@mstdn.social 2026-05-05 21:23

@w8emv@a2mi.social This is clearly damning of the readiness of uutils, but I think this is an opportunity that could also improve gnu-utils. I would like to know how many bugs have been found in gnu-utils as a result of trying to find bugs in uutils. The sudo-rs and sudo devs have been working together and finding bugs in sudo... even if setuid is deeply flawed, I see that as a unambiguously good. I don't blame the devs here, unless they have been pushing for uutils to replace gnu-utils. @hyc@mastodon.social @lcamtuf@infosec.exchange

Open ##2795521