|
|
|
|
|
by addaon
1301 days ago
|
|
Keep in mind that if you want to write probably-maybe-correct code, Rust is maturing to be able to get you there more easily than C. But if you want actually-correct code, you need to do the legwork regardless of language; and C has a much more mature ecosystem (things like CompCert C, etc) that lets you do much of the analysis portion of that legwork on C code, instead of on generated assembly code as you'd have to do for Rust. Combined with verification costs that don't vary that much from language to language, and there's a long future where, for safety-critical applications, there's no downside to C -- the cost of verification and analysis swamps the cost of writing the code, and the cost of qualifying a new language's toolchain would be absurd. For this reason, C has a long, long future as one of the few languages (along with Ada, where some folk are making a real investment in tool qualification) for critical code; and even if it takes a decade for C23 features to stabilize and make it to this population, well, we'll still be writing C code well beyond '33. |
|
If the language helps you avoid writing bugs in the first place, that should translate to quicker delivery and lower costs, as well as a reduced probability of bugs making it to production. The Ada folks are understandably keen to emphasise this in their promotional material.
As I understand it, this typically falls to the compiler vendor, not to the people who use the compiler. A compiler vendor targeting safety-critical applications will want to get their compiler certified, e.g. [0]. To my knowledge we're nowhere near a certified Rust compiler, although it seems some folks are trying. [1][0] https://www.ghs.com/products/compiler.html
[1] https://ferrous-systems.com/blog/sealed-rust-the-pitch/