Rust is opinionated. This isn’t a flaw — it’s the point.
The borrow checker, the mut keyword, the module system, the fact that cargo fmt has a default configuration that 90% of the ecosystem uses unchanged. Every opinion eliminates a class of bugs that C and C++ explicitly chose to allow because “the programmer knows what they’re doing.”
What “Opinionated” Means Here
C has exactly zero opinions about how you structure your code. You can:
- Put everything in a single translation unit
- Split across 200 files with no consistent pattern
- None of that matters to the compiler
Rust’s module system forces a tree. Macros are scoped. Traits require explicit implementation. Unsafe blocks must be annotated. None of these are arbitrary — they teach you to write code that the compiler can verify.
This is the fundamental trade-off: you give up “I know what I’m doing” freedom in exchange for “the compiler will catch your mistakes at compile time.”
The C Replacement Myth
Rust gets called a C replacement because it fills the same slot in the mental model: systems language, no GC, direct memory access, zero-cost abstractions. But this framing misses why C persists in the domains where it does.
C lives in three places:
-
Legacy codebases — Linux, OpenSSL, CPython. Rewriting these is a socio-technical problem, not a language problem. The code works, the maintenance teams know it, and the risk of subtle regressions outweighs the benefits.
-
Hardware-constrained environments — 8-bit microcontrollers with 2KB of RAM. Rust’s standard library doesn’t target these. LLVM backend doesn’t either. C wins because it’s the only practical option.
-
FFI boundary language — Every language can call C. This isn’t because C is good — it’s because C is the lowest common denominator. Rust is trying to become the new denominator via
extern "C"and#[no_mangle], but this takes decades, not years.
Rust replaces C best in the greenfield systems space: CLI tools, network services, parsers, databases, browsers (Servo), and OS components (Redox). These were C’s domains primarily due to historical accident, not technical necessity.
Is C++ the Real Target?
C++ is where Rust’s opinions hurt most. Template metaprogramming is replaced by generics + traits. Header files are gone. The build system is standardised. Object ownership is enforced at compile time rather than documented in comments.
The C++ ecosystem’s response to memory safety mandates (NSA, CISA, White House) tells you everything: rather than fix C++‘s footguns, they’re adopting Rust for new components. Microsoft, Google, and Amazon all do this in production today.
The Verdict
Rust will never fully replace C. Not because Rust is worse, but because C’s remaining strongholds either don’t benefit from Rust’s guarantees or can’t adopt Rust for practical reasons.
But Rust has already replaced C in the domains that matter for new systems development. The gradual migration of security-critical infrastructure from C/C++ to Rust isn’t hypothetical — it’s happening in the Linux kernel, Android, Firefox, and AWS.
Rust’s opinions make it harder to write bad code. That’s the whole value proposition. And for new projects where you control the stack, choosing C over Rust today is a bet that the borrow checker’s learning curve outweighs two decades of CVEs from use-after-free bugs.