[sunflowerdeath]

I think, it is very important, that people started to actually discuss and compare different approaches to safety, instead of just saying that since Rust is safer, we should throw out all of c/c++ code and rewrite everything in rust.

[com2kid]

It is amusing to me because last time I was using C, the problems rust solves weren't the problems I had in C.

Deeply embedded code doesn't use malloc, doesn't use threading.

I could use a better type system, ala Ada, being able to say "this variable is of type distance in meters, this variable is of type time in milliseconds", that'd have cut the # of bugs by a huge amount.

But simple, unsexy, type system changes like that aren't what language designers are focused on.

Who here has never confused Milliseconds and Seconds when passing a variable around? Trivial for a compiler to catch with a half decent type system, but few modern languages bother to try.

Even when writing modern code in newer languages, I rarely directly use threads, and if I need to pass data between them 95% of the time I can get away just doing a deep copy to avoid the hassles of sharing data between threads!

Obviously Rust is meant solving different problems than the ones I face, I have friends who frequently write highly threaded code, but in my day to day, Rust doesn't offer much more safety.

(However, Zig does look super cool and interesting!)

[asalahli]

Have you looked into the subset of D called "Better C"? I recently stumbled upon it and have been wanting to try it out. It seems to solve the exact same problem you're describing, though I don't know how good of a job it does at that.

https://dlang.org/spec/betterc.html

[steveklabnik]

I work on embedded code with no malloc, no threading. We still find Rust valuable. YMMV :)

[wjjvk]

This is so normal. You are one of the developers of rust. LoL

[steveklabnik]

The decision was made before I joined. I am also not the only developer.

[paragraft]

Re units of measurement in Rust, have you looked at something like https://docs.rs/uom/0.31.1/uom/ ?

[fulafel]

Oh we had this long before Rust, and most of C++ usage in new applications was displaced by safer (among other things) languages.

I think the biggest thing was that university curriculums and mainstream app development platforms (like Microsoft) stopped pushing it as hard when the level of horror got past a certain point. It used to be pretty bad. Business apps being written using MS "Active Template Library" in C++ and then used as signed ActiveX plugins on IE6-only web pages etc.

[smolder]

Safety (memory and otherwise) isn't new, but during my CS curriculum, including a course on programming language theory, there was little/no mention of techniques to ensure safety in the space between C++ and Java. I probably would have pointed toward formal verification if someone said they needed safety guarantees in the absence of garbage collection and a potentially slow or bloated runtime.

Though I believe there were some languages with features to that end, at least research languages, they weren't that well represented. I think Rust's presence brought attention to the possibilities there, and an increasing number of people see the value of investigating and developing that niche.

[pjmlp]

Microsoft still is the main company pushing it hard (C++ use) despite all security reports, most likely due to how the Windows and Office teams don't accept anything else.

So basically you have the DevTools and Azure teams pushing for .NET, Java and other safer languages, while Azure Sphere has a C only SDK and WinUI/UWP push C++ above anything else, with some C++ only APIs.

Politics.