What an absolute shitshow
Rust-rewriting is a kind of madness. I like Rust, it’s an amazing language. But why rewrite programs that existed for decades and have proven their stability and safety? Rewriting them to Rust won’t make them safer, it will just introduce the kind of issues original versions have got fixed long ago.
The MIT license also is a concern. I understand that many projects use it, and we can’t just reject them because of the license. But here we don’t see an innovation under MIT license - we see a copy of existing GNU tools, with hilarious issues and a corporation-friendly license.
The fact uutils are being shipped despite being so raw shows that this is not about better software. The whole project is about abolishing GPL. And Rust is just an excuse.
And the quality level of uutils being already shipped tells they either make free alpha testers for the corpos of the users, or there were no competent programmers to take part in the development.
C will remain the core of the modern digital world for many years. It is impossible to rewrite everything to Rust in a couple of years. It needs a careful professional approach if we really want this to make software better. But in this case, no one does.
But why rewrite programs that existed for decades and have proven their stability and safety? Rewriting them to Rust won’t make them safer, it will just introduce the kind of issues original versions have got fixed long ago.
Of course rewriting them will introduce some new issues, but it will also eliminate classes of bugs from which there are definitely still a great many in old “stable” C code (bugs which are now being discovered and will presumably continue to be discovered at a much faster pace due to LLMs).
The whole project is about abolishing GPL. And Rust is just an excuse.
I don’t think it is just an excuse; I believe that improving security is also a goal… but removing GPL code is clearly also part of their motivation :(
Eliminating some obscure bugs from C code is not worth intruducing a lot of new bugs. GNU coreutils have been used and polished for so long, that it would be far more effective just to fix the issues as they reveal right in the original code. If rewriting removes one kind of bugs while introducing another - then what’s the whole point?
I cannot imagine obviously buggy code from 2020s being more secure than code that has been around since previous century. Again, even if Rust for real is a better solution for security reasons, the way it is being developed and shipped is not how one makes software more secure. Disregarding the license, uutils look like something pursuing hype, not strategical benefits.
But bruh, “if it compiles, it works”. Who needs testing now that we have blazing safe rust with AI?
“Ship fast and break things”, bruh.
That’s the sad point where the software industry is at these days.
In a few years people will be locked-in with some proprietary Linux distro variants made by big tech and they will wonder how that happened.
People show stop a moment and reflect on why the GNU license exists in the first place.
Seeing what bun did, maybe all of that C tooling is just a weekend with Claude away from being ported.
Not saying that is a good thing, I don’t support that, but what is stopping stupid people from doing that.
I just started a hobby project with Bun, it’s my first time trying it. I would like to know what you are referring to when you say “what Bun did”. Can you share more information?
He’s talking about Bun being completely rewritten from Zig to Rust by Claude after the project was acquired by Anthropic
Thanks, I didn’t know any of this.
This can’t be legally done with a GNU project, since this is a direct derivation. I also don’t know if bun is ok now and if it used to be before.
Holy crap.
I love rust, but I absolutely hate how it’s used to jam MIT licenses where GPL belongs. Maybe it’s time we consider using corpos tools against them, and use an AI to rewrite GNU utils to Rust, so that people can continue contributing to Rust while not feeding corps?
Edit: Though licensing AI software is iffy at best, you’ve got to own the copyright to something to licence it: Non-human productions are legally non-copyrightable. Also it might be better to just have humans do it anyway. The intent of my message was just that maybe we ought to deprive MIT-licensed projects from FOSS-motivated developers by providing Rust GPL alternatives to MIT/corporate Rust projects
Isn’t the MIT license independent of the choice of Rust?
Rust often ends up just being an excuse to rewrite software with corporate-friendly licenses without copyleft. That’s not necessarily true though, Lemmy itself is Rust & licenced under AGPL
You can use rust and still use the GPL.
My issue isn’t with Rust as a language at all, I quite enjoy making my projects with it. My issue is with “Rust rewrites” of GPL software, only to have those rewrites be licensed under MIT/Apache. To me it signifies that these rewrites were never about the safety features of Rust, but that they are attempts at pushing out the GPL
The real shitshow is it’s MIT licensing. Corporate takeover 101
Corporate takeover?
Ubuntu has always belong to a corporation.
It’s not like it’s a community project. Ubuntu has always belonged to Canonical.
Coreutils isn’t only for ubuntu. Ubuntu just seems to be the lab rat corporation.
There are many kinds of corporations, my friend. Canonical is different from Apple. Wait a second… Wasn’t OS X built on FreeBSD?
How is it a takeover, if they write their own set of tools?
The gnu brained folks hate when we make our own tools.
Square that circle for us?
Have you ever publish a tool that already has a gnu counterpart? Even if you say it’s for learning or an experiment you still get hounded about it.
Why publish an experiment?
It’s my git server, I can put what I want on it. MIT is easy when I don’t really have long term plans for it. If I am contributing to a GPL project i’ll use their license.
To get feedback? For fun?
I can see that, I guess even practicing deployment would be useful. There’s enough weird splintering in some projects that even perceived competition can seem like trouble.
What an absolute shitshow
I’d say the month of June is actually a good time to be breaking and fixing things in a release that is due to come out in (checks notes) October.
No, it is not a good time. A project like Ubuntu should now be in freeze as they had about 3 months before release and definitely it should not have a break in something basic just because the language used to write the command break backword compatibilty
No, it is not a good time. A project like Ubuntu should now be in freeze as they had about 3 months before release
This bug was reported (and resolved by rolling back to the GNU coreutils version of
cp) on June 30, a little over 15 weeks prior to the scheduled release date.Which distros have a feature freeze that far in advance?
Ubuntu hasn’t even scheduled theirs for 26.10 yet; if you edit that url to look at previous releases’ schedules you can see their feature freeze and debian import freezes are typically about 2 months prior to release. (See here for descriptions of all of the different types of freezes…)
I like staying up to date about open source but holy cow is there too many of these “omg they broke something in testing”. Yah, that’s the point.
The project hasn’t had a stable release, and yes, it does certainly need more testing to uncover edge cases.
Yes, MIT bad, but one must not diss on the project just because it has been written in Rust.
The problem isn’t the language. It’s the cargo cult that surrounds it.
I see what you did there
I disagree with MIT License being bad. I agree on all other fronts of your statements.
Why don’t you think the MIT license is bad?
Why do you think the MIT License is bad? I am not the one making the claim it being bad, so I’m not the on in defending position. It’s an open source license and I like to use it too (granted my work is just little small hobby tools). I think the MIT License has pros and cons, but isn’t straight a bad license in this context.
I guess it makes coreutils vulnerable to EEE. Corps with significant stakes in Linux would be able to extend the API with proprietary functions, build ecosystems on top of that, and lock out independent development.
Imagine being unable to distribute a Linux based OS that is compatible with most of the world without proprietary age verification built in for example.
If its proprietary, then it can’t be distributed in an Open Source system anyway. We are free to use the Open Source version instead. I’m actually in favor of companies having the option to do what they want with the code, that is true Open Source in my opinion. In example a different distribution or even Microsoft takes the coreutils, and makes changes and then distributes it as proprietary software. We wouldn’t need to use it anyway.
Also which MIT Licensend software happened to this in our Linux eco system, that it became to a problem? Are there important examples?
Which also means Microsoft™ coreutils will always be “better” for general public. Yeah you are talking about open source, but this is more about Free Software ideology
Who says it will be better? It might also be worse. It might be bloated with functionality we don’t care, it might get telemetry. If its proprietary, we can’t trust it anymore. I don’t think this would be “better” just because Microsoft did something add proprietary features to it. Even the general public don’t think that Microsoft software is the best software.
Besides that, just because Microsoft does it better does not mean ours would get worse. It wouldn’t affect Linux at all.
MIT is terrible if it replaces current GPL projects. Companies will always provide their spyware infested proprietary version of the exact same thing which have one or two additional features, making open source software always behind rhe propruetary counterparts. See: Chromium->Google Chrome, Aosp->any Android os vendor
Great projects have used MIT without any issues. Godot for instance, which may also be needed, I don’t know if games made by godot could be closed source if it would use gnu license for instance.
Godot didn’t replace an existing broadly used GPL game engine, so this is irrelevant.
The logic was that with a mit license companies will provide a copy of the software infected with spyware leaving the open source project behind.
Explain why that hasn’t happened to godot.
No, the logic was that replacing a GPL project with an MIT project is bad.
Because Godot is already quite ibferior to its proprietary alternatives, atleast in popularity. If godot was The game engine that everyone uses, proprietary ones will come and try to have it. They can have all the godot features as well as something new from their side
Could you provide some examples when that had happened?
I’m looking up famous projects using mit license and in any of those that had happened.
Lua, node.js, jQuery…
Even X11 which was indeed replaced by other system… Wayland, which also uses MIT license.
The majority of project are MIT licensed and it’s not even close.
That cp went really hard apparently!
People will blame Rust for the incompetence of Ubuntu team to adopt the uutils as default prematurely.
they broke something in testing. that’s not incompetence, that’s the whole point.
I will say the Rust stdlib seems to make TOCTOU bugs really easy to make for filesystem operations
But, yes, Ubuntu switching to a test project hurts it
Im so happy work stopped using ubuntu server after last time.
Also how the heck do you break cp of all things.
I dont want to experiment with core utils.
Give me 10 minutes and I’ll write you a cp that is completely fucked.
cp $1 $2
I invite you to peruse the man page for cp.
cp of all things
cp might sound simple because its a very necessary thing for an OS to do, but there’s quite some technical depth to each of the core utils, if it were simple people would just be pumping out coreutil practice projects just like they do with “generic CRUD web app 5000”
if it were simple people would just be pumping out coreutil practice projects
you might be surprised how many results come up when you search for this sort of thing
When i’m in the most unnecessary competition and my opponent is rust coreutils:
Memory safety anyone?
GNU Coreutils have worked very well for so long so replacing it is totally pointless unless memory safety has crippled the project somehow till now.
Also I really hate that project for another reason - License
Yep. And more distros support so theres more incentive to fix any issues that have popped up. Plus theres more c devs in the world than rust.
Rust is faster and easier to develop in than C. It’s also safer.
What’s wrong with the uutils license?
Rust is absolutely not faster or easier than C. It’s safer but that’s it.
The issue with the rust utils license is that its MIT. Which many people see as vastly inferior to GPL. This is a very big deal to people who believe in the GPL.
It is definitely easier.
Don’t try and tell me cargo is not easier than fucking around with the C/CPP build slop I’ve had to screw around with over the years.
The coreutils license is bullshit but let’s not pretend languages have not improved in 50 years.
Yeah, the only real issue I have with the “rewrite it in Rust” approach is the absolute plague of permissive licensing. It is much easier to write safe, correct code in Rust.
I think it depends on the project, when writing cp you’re not really messing around with libraries so good ol make is fine
Cargo != Rust
There is no word in which the borrow checker is easier to deal with than C.
I think responding to borrow checker errors is easier than debugging some random segfault
Also I don’t separate the language and it’s build system. I have to use it to write the language it’s an entire kit.
Part of c/cpp problem is it’s build system and dependency story just serves to get in my way instead of letting me write code. The syntax is irrelevant when I’m fucking with some auto conf script because the version on my machine is 0.01 versions different.
Rust is absolutely not faster or easier than C. It’s safer but that’s it.
Depends on what you mean by “easier”. It is easier to write safe code in Rust.
Memory safe!
Not any other kind of safety.
Yes, that’s what Rust is about.
Depends on what you do, really.
Dangers of this project:
- compatibility in edge-cases
- experts of their tool vs. jokel of all
- maturity of the code
- scope creep
No, it does not depend on what I do. In Rust it is by definition easier to write safe code than in C.
GNU coreutils are a copy of Unix utils themselves.
They’re also free and open source, which unix wasn’t.
Which is against the copyright in a same way as rust rewrite is
Rust is web scale!
