Who is NOT moving from C/C++ to Rust? | Hacker News Reader

Who is NOT moving from C/C++ to Rust?

5 pointsmbn859 comments

There is a clear trend of companies interested in migrating existing C/C++ to Rust, with notable examples like Microsoft, ANSSI, DARPA, etc.

Do we also have examples of companies that, after thoughtful consideration, decided that this should NOT be pursued in the foreseeable future? (ie. next 10y)

Since we usually see pro-rust or neutral/maybe positioning (eg. Linux kernel), it would be interesting to know the counter arguments.

PS: In the context of large projects, with long-term support in mind, and that decided to stay on C/C++ specifically -- not that Rust isn't chosen but Zig/other is instead).

Best

View on HN

I work in a sector that manufactures industrial control equipment (this is cutting-edge stuff). As far as I'm aware, nobody is seriously considering moving away from C/C++ at this time. The advantages to doing so do not justify the risks and expense.

Speaking of Linux Kernel - how much Rust has made it into the source? I think close to zero. Most of the companies that are 'interested' are just doing marketing gimmicks to attract younger talent but the reality is most of the work goes on in C++ and some edge cases are done in Rust.

My company works in high performance computing and there is no reason to move to Rust in the next 10 years.

It is hard to judge about companies which are naturally sharing some anti-academic attitude, so it is more decent to analyze some information from FOSS movement. It quickly becomes noticeable that there are no such language as "C/C++" and one of my favorite writings in the field of C programming language is [1]. I believe most software of GNU movement keeps using C as their base language until something more persuading will be shipped and I consider both Rust and C++ as not that beautiful C replacements because of too much overhead.

For example, neither PL from three mentioned ones can not do templates which are not a PITA when dealing with more than one data type. IMO (inspired by Lisp of course) this feature is more important than RAII or borrowing which effectively works only with tree-shaped data structures.

[1] http://250bpm.com/blog:4

I currently work on certified avionics. I don't see rust used there in the next 10 years.

Even using OO over plain C adds a huge amount of extra work already.

>> I currently work on certified avionics. I don't see rust used there in the next 10 years.

There is definitely interest in using Rust for safety critical work:

https://www.adacore.com/gnatpro-rust

https://ferrocene.dev/en/

I think it will definitely happen, but I am not sure how soon.

thesuperbigfrog same references :P That's what I call independent analysis, lol

Could you please elaborate on why?

Projects like Ferrocene or GNAT Pro seem to indicate that's a possibility. (automotive gaining traction sort of paves the way to avionics/aerospace).

https://ferrocene.dev/en/ https://www.eenewseurope.com/en/first-rust-compiler-qualifie...

I don't see any incentive to move away from C. I was discouraged to use C++ even. The amount of extra work is huge.

I don't know a lot about Rust myself, but the impression I get is that it is a bit less obviously deterministic than C.

Things boil down to (IMHO): - every requirement must be traceable to code - every line of code must fulfill a requirement - any dynamic behavior (for example heap allocation) must be proven to work under all circumstances

Large game publisher shipping console/pc games in custom engines. Even if someone is ninja coding new engine in Rust right now I am likely to retire before it will be in use. Ofc I can not say for whole company if it was decided to not use Rust or not.

Reply on news.ycombinator.com