Does this fix something specific with coreutils (is there no way to upstream patches?)? Or is it that rust is supposed to save us from some class of unknowns (related to memory safety?)?
The world is slowly but steadily going to replace non memory safe code by memory safe code (no matter how people feel about it). This is just part of that.
Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
11–20 of 24 posts
Re: Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
#12That's the worst idea I've seen in a while. People rely on the behaviors of the GNU coreutils, this is going to break a lot of stuff. The uutils are not there yet at all.
Re: Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
#13> The replacement will affect more than a hundred utilities that are part of Coreutils … > Last week, the issue of the package uutils coreutils 0.0.30 successfully passed 507 tests (in the last issue of 506, in a year old – 476) from the standard GNU Coreutils test set. If I’m reading this right then there are on average only one to five tests per utility (like cat, dd, …) which is a shockingly low number. Particular…
Re: Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
#14Does GNU coreutils have that many CVE?
The question is rather how many are logic bugs and how many are memory safety bugs, Rust excludes the latter quite well, which according to the MS and Google Studies makes up about 70% of programming mistakes in their C++ codebases. However rewriting carries the risk of introducing new logic bugs. So There's a tradeoff here.
I am a big believer in Rust, I basically only use rg instead of grep and fd instead of find and never looked back, but they are slightly different in behavior, so not a drop-in replacement. I'm not sure if porting coreutils will be so useful in the end, since if you do all the workarounds for the rust version to behave like the C version, maybe you won't have the performance benefits or the clean codebase that you set out for in the end. I like the rg and fd approach much better. It's an almost drop-in with huge performance benefits.
But let's see if they succeed.
Re: Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
#15Earlier quoted context omitted.
The world is slowly but steadily going to replace non memory safe code by memory safe code (no matter how people feel about it). This is just part of that.
It seems like replacing code that has been written and used extensively for decades with a complete rewrite without a real need is more of a risk to me. Perhaps the uutils are 'simple' enough not to be an issue though.
A lot of code will need to be replaced by memory safe code in the future, but a lot of code is not so easy to replace as something like this.
Re: Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
#16Does GNU coreutils have that many CVE?
I wouldn't say many, but not none: https://app.opencve.io/cve/?vendor=gnu&product=coreutils The question is rather how many are logic bugs and how many are memory safety bugs, Rust excludes the latter quite well, which according to the MS and Google Studies makes up about 70% of programming mistakes in their C++ codebases. However rewriting carries the risk of introducing new logic bugs. So There's a tradeoff here. I…
Re: Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
#17That's the worst idea I've seen in a while. People rely on the behaviors of the GNU coreutils, this is going to break a lot of stuff. The uutils are not there yet at all.
I think this is a good experiment. 25.10 is not an LTS, people who want stability are using LTS anyway.
Re: Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
#18> The replacement will affect more than a hundred utilities that are part of Coreutils … > Last week, the issue of the package uutils coreutils 0.0.30 successfully passed 507 tests (in the last issue of 506, in a year old – 476) from the standard GNU Coreutils test set. If I’m reading this right then there are on average only one to five tests per utility (like cat, dd, …) which is a shockingly low number. Particular…
Many of those were written in 80s, testing was not a thing.
Re: Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
#19Earlier quoted context omitted.
I wouldn't say many, but not none: https://app.opencve.io/cve/?vendor=gnu&product=coreutils The question is rather how many are logic bugs and how many are memory safety bugs, Rust excludes the latter quite well, which according to the MS and Google Studies makes up about 70% of programming mistakes in their C++ codebases. However rewriting carries the risk of introducing new logic bugs. So There's a tradeoff here. I…
Do you have a source for “70% of programming mistakes ..” part? I was looking for this but couldn’t find it
Re: Ubuntu 25.10 Replaces GNU Coreutils with Rust Uutils
#20That's the worst idea I've seen in a while. People rely on the behaviors of the GNU coreutils, this is going to break a lot of stuff. The uutils are not there yet at all.
"Differences with GNU are treated as bugs." - https://github.com/uutils/coreutils#goals
That's promising. Wouldn't it be annoying though, to have to keep things bug-for-bug compatible on a green-field project? That priority may be the first compromise, and divergence is inevitable.
(If few report edge cases and there's no fuzzing.)
[1] https://github.com/coreutils/coreutils?tab=GPL-3.0-1-ov-file