Hacker News Reader

Comment by eikenberry | Hacker News Reader

Have you looked at Zig? It is often termed a modern C where Rust is the modern C++. Seems like a good fit.

i never really understand why these get compared. i wouldn't expect that much overlap in the audiences

zig seems like someone wanted something between C and "the good parts" of C++, with the generations of cruft scrubbed out

rust seems like someone wanted a haskell-flavoured replacement for C++, and memory-safety

i would expect "zig for C++" to look more like D or Carbon than rust. and i'd expect "rust for C" to have memory safety and regions, and probably steal a few ocaml features

The single best feature (and I would say the _core_ feature separating it from C) that C++ has to offer is RAII and zig does not have that. So I don’t know which good parts of C++ they kept. Zig is more of its own thing, and they take from wherever they like, not just C++.

Zig has defer, which is arguably way simpler. Is there something RAII can do that defer can't?

Not everyone likes RAII by itself. Allocating and deallocating things one at a time is not always efficient. That is not the only way to use RAII but it's the most prevalent way.

defer can't emulate the destructors of objects that outlives their lexical scope. Return values and heap objects are examples of these since they outlive the function they were created in. defer only supports enqueuing actions at lexical boundaries.

If you destroy an object that outlives the lexical scope it was created in, then you have to clean up manually.

Defer can be omitted

> So I don’t know which good parts of C++ they kept.

comptime is a better version of C++ templates.

I really wish that were true but it isn’t. Modern C++ templates/constexpr are much more powerful and expressive than any Zig comptime equivalent.

The power and expressiveness of the C++ compile-time capabilities are the one thing I strongly miss when using other languages. The amount of safety and conciseness those features enable makes not having them feel like a giant step backward. Honestly, if another systems language had something of similar capability I’d consider switching.

I have written a lot of Zig comptime code and ended up finding the opposite. In C++ I find I have to bend over backward to get what I want done, often resulting in insane compile times. I've used metaprogramming libraries like Boost Hana before to have some more ergonomics, but even that I would consider inferior to comptime.

Out of curiosity, do you happen to have any examples of what you describe, where C++ is more powerful and expressive than Zig?

If it looks anything like what I read in "Modern C++ Design" 20+ years ago then I'll pass. That book made me realize the language wasn't for me anymore.

It looks nothing like C++ decades ago, it is effectively a completely different language. I found C++ unusable before C++11, and even C++11 feels archaic these days. Idiomatic C++20 and later is almost a decent language.

"Modern C++ Design" was the first book that highlighted template metaprogramming and showed you ways to use it as a Turing complete programming language in itself. Even a decade+ after it was printed my friend said it was recommended reading for employees within Google so I gave him mine - but I'm not sure about these days.

Talking about how much you can do with C++ templates made me think of that.

It’s not like that anymore.

I would say OCaml more than Haskell, but yes.

I have, and I do find Zig impressive, but it doesn't go far enough for me. I don't want a "better C", I want a better systems language that can also scale up for other uses.

I like strong, featureful type systems and functional programming; Zig doesn't really fit the bill for me there. Rust is missing a few things I want (like higher-kinded types; GATs don't go far enough for me), but it's incredible how they've managed to build so many zero- and low-cost abstractions and make Rust feel like quite a high-level language sometimes.

Almost everyone wants to claim to be modern c and not modern C++ because everyone shit on C++

If anyone is the modern c++ its D. They have template metaprogramming while rust has generics and macros.

Rust doesn't have constructors or classes or overloading. I believe its type system is based on Hindley–Milner like ML or Haskell and traits are similar to Haskell type classes. Enums are like tagged Union/sum types in functional programming and rust uses Error/Option types like them. I believe rust macros were inspired by Scheme. And finally a lot of what makes it unique was inspired by Cyclone (an obscure excitemenal language that tried to be a safer C) and other obscure research languages.

I guess rust has RAII that's one major similar to c++. An there's probably some similarities in low level memory access abstraction patterns.

But I'd describe rust as more of an imperative non GC offshoot of MLs then a modern c++ evolution.

Rust is not a modern C++, their core models are pretty different. Both Rust and C++ can do things the other can’t do. C++ is a more focused on low-level hyper-optimized systems programming, Rust is a bit higher level and has stronger guardrails but with performance closer to a classic systems language.

I do think Zig is a worthy successor to C and isn’t trying to be C++. I programmed in C for a long time and Zig has a long list of sensible features I wish C had back then. If C had been like Zig I might never have left C.

What do you consider the difference in their core models.

Rust and C++ both use RAII, both have a strong emphasis on type safety, Rust just takes that to the extreme.

I would like to even hope both believe in 0 cost abstractions, which contrary to popular belief isn't no cost, but no cost over doing the same thing yourself.

In many cases it's not even 0 cost, it's negative cost since using declarative programming can allow the compiler to optimise in ways you don't know about.

Rust is basically what you get if you start with ML and then do enough to avoid having garbage collection.

This seems not correct at all. There are very useful ML patterns that don't exist at all.

Rust feels like taking an imperative language and adding ML features till you cannot anymore.

C++ is (according to Bjarne Stroustrup) a general purpose programming language that can be used to build general business software and applications with, not just a systems PL. This is why perf is all over the place —- the core language is fast like C but the stdlib contains terribly slow code (regex, exceptions) and ways to waste lots of cycles (wrapping everything in smart pointers, using std::map instead of unordered map).

What's slow about regex? A properly optimised regex engine ought to be very efficient (if used sensibly, at least).

std::regex in C++ cannot be "properly optimized" due to various factors: https://stackoverflow.com/questions/70583395/why-is-stdregex...

The answer there [0] doesn't support this. From the answer:

> There is no question that libstdc++'s implementation of <regex> is not well optimized. But there is more to it than that. It's not that the standard requirements inhibit optimizations so much as the standard requirements inhibit changes.

The answer's one comment expands on this, it sounds like they're not able to add a sophisticated optimising regex engine into the libstd++ shared library (i.e. non-inline) as this would be an ABI break.

Perhaps other implementations of the C++ standard library perform better.

[0] https://stackoverflow.com/a/70587711/

In my understanding, other implementations cannot perform better, because the root cause is how it is defined in the standard. Basically, any better implementation would be non-conforming. It’s not the kind of ABI issue where simply choosing a different one would help; the flaw comes directly from the definition, which cannot be changed.

Of course, non-std implementations of regexes in C++ don’t have this issue. Were strictly talking about the standard library one.

Reply on news.ycombinator.com