Unlike you babies I have Personal Responsibility and I write all of my code directly in assembly the way reagan intended. I don't need guard rails and I've never had any issues with it because my Personal Responsibility keeps me safe
post
C's compiler prevents common type bugs and handles things like register allocation for you? So does skill.
I love this argument because it means this dude is the only skilled C developer on the planet. Chromium devs are just chumps that should be replaced by this uncommon God.
"So does skill" I agree 100%
However, we're human. You show me a skilled developer who never causes bugs, and I'll show you a liar.
No matter how skilled or experienced a developer is, they always have the capacity to introduce a bug by accident.
Whether it's a typo, or simply being tired or distracted, or just having one of those moments, or even one of those days. It's completely normal.
Coding is just communication, and when working on larger codebases it can be just as nuanced as interpersonal communication. People miscommunicate every second of the day.
I've never used Rust.
“Should I use rust or c++” is the wrong question IMO. The right question is “do I want the code I run, written by thousands or millions of randos, to be written in rust or c++”.
The human mind has limited capacity for things to pay attention to. If your attention is occupied with tiptoeing around the loaded guns scattered all over the floor, sooner or later you’ll slip and trip over one.
Of course, you’re a virtuoso programmer, so you can pirouette balletically around the floorguns as you deliver brilliantly efficient code. Which is great, until you have an off day, or you get bored of coding, run off to join the circus as a professional knife-juggler and your codebase is inherited by someone of more conventional aptitude.
Programming languages offering to keep track of some of the things programmers need to be aware of has been a boon for maintainability of code and, yes, security. Like type systems: there’s a reason we no longer write assembly language, squeezing multiple things into the bits of a register, unless we’re doing party tricks like demo coding or trying to push very limited systems to their limits.
... until you have an off day, or you get bored of coding, run off to join the circus as a professional knife-juggler and your codebase is inherited by someone of more conventional aptitude.
Sometimes you even have to deal with having mere mortals on your team!
Which is why garbage collection is the way to go.
It is a way to go but there are still cons there. Guaranteeing memory safety isn't free. You have to pay for it somewhere, either at compilation time, like Rust, or during runtime like in Go. Both are solid approaches but GC will cause problems in cases where the extra runtime overhead is not acceptable.
- if your skill is so great that you would never cause the kinds of bugs the rust compiler is designed to prevent, then it will never keep you from compiling, and therefore your complaint is unnecessary and you can happily use rust
- if you do encounter these error messages, then you are apparently not skilled enough to not use rust, and should use rust
In summary: use rust.
Your first point is not true. There are valid uses of memory sharing that rust will reject.
Curious what you are talking about. Multi-threaded sharing of memory for example is also easy with rust, it just doesn't let you wrote and read at the same time, and so on.
Off the top of my head, single-threaded writing to the same memory from different fields of a struct. Not to mention self-referencing like if you want to hold a buffer and have different views into it in the same structure.
Classic example: A linked list
... are entirely possible, even if rarely the right choice.
This so true, every one complaining that the borrow checker is annoying isn't apparently aware what they used to do was inherently flawed. Sure there a some, though rare, false positives. But they are easily mitigated. These people are exactly that what they themselves are complaining about, elitist.
Skill issue takes are dumb as fuck. It's just republican personal responsibility takes using different language.
Intelligent people focus on producing systemically better outcomes.
While I do totally see the advantages of rust and agree skill is not a solution given people make mistakes...I do agree a lot of the very vocal rust advocates do act almost religious and it is an annoying turn off.
We had the Java guys in year 2000, at least Rust seems to be a decent language.
Java was created so that teams of intermediate skill programmers could maintain large, long-lived code bases. And it did its job incredibly well.
If that is not your use case (or you do not want to admit that you are such a programmer), it may not be your favourite language.
I always like C# far better. It may be my favourite language overall. It has a bit more headroom and was designed somebody far more skilled. But it was designed to compete directly with Java. So, you know who it was built for.
I've never run into a Java evangelist. Every opinion I've ever heard about Java is something like "Yeah, this sucks". I always thought that people put up with it because it's write-once, run-anywhere, but so is, y'know, Python.
Early on, Java was advertized as the next great thing, ending headaches from system development, porting, and "promoting good programming practices through OOP".
Then people increasingly got tired of OOP and the speed penalty of both that paradigm and the JVM, not to mention more and more education institutes started to claim Java was too hard for beginners, and that Python would be better.
Now we have Rust evangelists promoting the language as the next great thing, ending headaches from memory safety issues, porting (if you target WASM and pack your app into a Chromium instance), and "promoting good programming practices through FP".
Time is truly a flat circle...
There was a saying back in the day, roughly: "java can run on all platforms like anal sex works on all genders".
Python is slow but fantastic when it comes to interoperability IMO and is just complex enough that you can get the job done. I just hope they'll won't complexify it into oblivion, it's a really neat language. IMO.
Do you have time to talk about our lord Rust? Did you know it died for our bugs so we don't have to debug them at run time?
The language/compiler restricting some bad practices makes safer code with less skill. Btw, there's also Safe C++
I can sympathize with some people getting tired of "rewrite it in Rust", especially when it's suggested inappropriately. (Worst I saw was an issue opened on something, maybe a database, I don't remember. Someone said they were a new programmer and wanted to help and only knew a little Rust and that if the project was rewritten in Rust they could help.) But... Rust's compiler being able to do those things is actually super useful and amazing. This is like someone saying they don't need static types because they know the language good enough to not misuse the dynamic types. This is like someone saying they don't need C because they're good at assembly.
While it isn't something as simple as Rust being strictly better than C/C++, it's really silly to say that you being a good developer means you don't need guardrails. Everybody makes mistakes all the time. You're not perfect.
I'd guess it's Rust fan's genuine belief that they have something revolutionary.
“Rust’s compiler prevents common bugs” So does skill. No offense to you, but, this trope is getting so tiresome. If you like the language then go ahead and use it.
If you're that much of a galaxybrain, you should be writing everything directly in opcodes. In reality, nobody is, and we invented languages to help us perform an activity the human brain is very poorly suited to.
This attitude also means that OP stares at their own obvious bugs on a screen all day and then decides they're great, which is level of detachment from reality frightening to me.
Sadly, it is a detachment from reality that is entirely normal, even typical. In all walks of life.
What I still find surprising, even though normal, is how technical people can push actual facts and evidence right out of their world view.
Sure, 70% of the bugs in C++ code bases are memory rated according to multiple sources. So let me aggressively and confidently berate this idiot that says the Rust compiler is doing something useful.
You do not have to use either language to see how idiotic this is. Even if you accept that this guy has “the skill” to make compiler help redundant, he has no point at all unless he thinks that “typical” C++ users have that same level of skill. And, provably and trivially researched—they do not. Being this wrong makes him, as self-evidenced, incompetent by definition.
All he proves in the end is that he is angry (and I guess not a fan of Rust).
“Angry and incompetent” is sadly a much more common trope than the ones he tires off.
There's some weird effects with language-specific bug rates.
In old Java, most uncaught exceptions are NullpointerExceptions, because most other exceptions used to be checked. Can't not catch a checked exception.
So they made Kotlin, where NullpointerExceptions are the only type of checked exceptions. Now there are no unhandled NPEs anymore but now you get tons of other exceptions.
Oh yes, it's so very human nature. But damn.
Most coders get the message at least a bit, I think. Other engineers have a reputation for massive egotism, software engineers don't really.
Other engineers have a reputation for massive egotism, software engineers don’t really.
That's a joke right?
Well, it's possible I'm missing something, or that there's a different reputation actually in the industry, since I'm an amateur. The first stereotypes I think of are unkempt, caffeine-dependent and socially inept.
When I've seen people asking for help online, traditional engineers seem much more likely to flex their credentials and then not actually answer. Although there's definitely software examples as well.
Because most projects are worked on by multiple people, and you shouldn't trust that everyone who will work on something will have the same skill level as you
If there are two languages otherwise equivalent in NFRs, where one lets you make the mistakes and the other doesn't, you're a bit silly if you don't pick the latter.
Good engineers shouldn't struggle to use a different language, so that's not an argument
I mean... they do kinda have a point on the last part. I'm no programmer or coder. I can't code for shit. I don't know a lot about development. And even I have the feeling that Rust people, they're kinda like NixOS people a while back, they never shut the fuck up about it. :3
They're definitely enthusiastic, I'll give them that. But so many projects are sold solely on the fact that they are made with Rust, even though it means absolutely nothing to most users.
I remember when System76 announced that they were making a new desktop environment and the only thing they basically said about it back then was that it was made with Rust and it felt like my corner of the internet lost their mind about it like they had announced the second coming of Christ or something.
Gonna guess, basically the same thing. Easy answers to hard questions instead of you having to think about them.
So, as far as they would be concerned, the only reason more people haven't chosen that path must be because they don't know how much easier it is, and how much less they have to think about stuff.
They can't see that building skill and knowledge has value beyond the extra effort.
At my last job I worked in a code base written in C and it needed to be certified to MISRA level A, and even in a language with as many foot guns as C, it's possible to write safe code. You just need to know what you're doing. I know there are tons of Rust zealots out there claiming it'll solve every last problem, but it turns out you just need to be careful.
The "common bugs" that the Rust compiler prevents are those a good programmer should not make in the first place. It's the bugs that even evade a seasoned programmer that poses the problems, and there, Rust won't help either.
Remember ADA? A programming language frankesteined by a committee to make programming safer? The programmers using it still produce bugs. And ADA is way more whips and chains than Rust.
It's the bugs that even evade a seasoned programmer that poses the problems, and there, Rust won't help either
What do you mean these are not the ones that rust tries to fix? Even huge projects like the linux kernel get memory bugs. I don't know anything about ADA and nor do I want to "evangelize rust" but what you're saying sounds boggers.
Obviously rust cannot prevent all bugs or even most of them. It can only prevent a small subset of bugs, but saying that that "small subset of bugs" wouldn't happen to seasoned programmers is just wrong, especially when you have tons of programmers working on the same big project.
I don't mean to say that rust is always the correct choice, but that you're waving off its greatest offering as bicycle training wheels (i.e. something no seasoned programmer would need)
but what you’re saying sounds boggers.
Believe me, it isn't. I program about anything for forty+ years now. I probably have forgotten more programming languages than you can list, and if there are constants in programming, then a) while compilers get better at catching bugs, they never got over the basics, and b) a good programmer will alyways be better at preventing and catching bugs than a compiler.
Once you have aquired a good mindset about disciplined programming, those buglets a compiler (or even code review systems) can find usually don't happen. Be wary of those bugs that evade the seasoned programmer, though...
For the mindset, it is good to read and understand things like the MISRA standard. Stuff like that prevents bugs.
Man hackers would love you. Memory vulnerabilities are still one of if not the main cause of vulnerabilities in unsafe languages. Yes this even applies to large projects ran by experienced and skilled people like the Linux kernel. Stop thinking your the exception, because I guarantee you aren't but even if you were you can't read all of the code in the world in your lifetime. Nor can you write all the code in the world.
I probably have forgotten more programming languages than you can list, and if there are constants in programming, then a) while compilers get better at catching bugs, they never got over the basics, and b) a good programmer will alyways be better at preventing and catching bugs than a compiler.
I agree with this
Once you have aquired a good mindset about disciplined programming, those buglets a compiler (or even code review systems) can find usually don't happen.
I also agree with this.
I would like to put a lot of emphasis in the usually. It doesn't mean that they don't happen, no human being makes no mistakes. Rust simply gives people a little more peace of mind knowing that unless they use unsafe they're probably fine in terms of memory issues.
As a side note, there was this once I was making an ecs engine in rust, and kept fighting the compiler on this issue. Specifically, the game engine bevy uses Query in the World to retrieve information about the game state, and I wanted to do the same. For instance, in the following function (or something similar, I honestly don't remember all that well):
fn getplayer(player: Query<Player>) {}
Would get player from the world and assign it to player (more or less). However rust was adamant in not letting me do this. After some thinking I finally realized why
fn getplayer(player: Query<Player>, player_too: Query<Player>) {}
Would give two mutable references to the same Player in the same function, which can be very easily mishandled, and thus is not allowed in rust.
I don't know about the MISRA standard, but I don't think that using it would have changed the way I coded my inherently flawed approach. This is a small example, one that didn't even matter that much in the grand scheme of things and could be even hard to understand why it's bad without knowing rust, but it is the one that came to mind. I think that if I had more experience I would he able to give you one that actually had security implications.
I'm no seasoned programmer, however
This is the same kind of argument that you shouldn't have to wear seat belts because just don't crash, duh.
Why do we have traffic lights?? Just don't hit each other, duh.
You shouldn't need emergency shutoffs at gas stations, don't have a fire stupid!
Hand rails on stairs? C'mon, just don't trip.
Confirmarion dialog on deleting data? Well, just don't click the wrong thing 🤦
It's beyond obtuse
And completely missed the entire problem of human error. Which is one of the only constants we can always rely upon in every situation in every industry everywhere.
Being ignorant of something this pervasive and uniform across all humanity is not an excuse.
all 55 comments