Hacker News

points by devsda 4 days ago | 0 comments

I read that as "is c++ still relevant for new features, given Rust and its popularity", not "who is still using C++".

Personally, I think it is still relevant. Many people thought Kotlin could overtake Java, Java reinvented itself with new initiatives/features and was saved from becoming irrelevant.

Rust also has the LLM advantage which is where I think the "if it compiles it works" nature of Rust is easier for manual changes and maintenance when needed.

[–] afdbcreid 4 days ago

Java vs. Kotlin is a different matter. It's still an open question if C++ even could match Rust's features, not to say if it (the committee) wants to, and the answer seems to be "no".

[–] gmueckl 4 days ago

The feature parity question goes both ways: rust won't ever be able to express some things that are possible in C++.

[–] afdbcreid 4 days ago

Like?

[–] gmueckl 4 days ago

Algorithms that safely mutate the same data structure on multiple threads and do so safely because of some complex algorithmic properties that guarantee that the changes do not overlap?

[–] afdbcreid 4 days ago

Rust can do that just fine. https://doc.rust-lang.org/stable/std/cell/struct.UnsafeCell..... "some complex algorithmic properties that guarantee that the changes do not overlap" are exactly the place to use unsafe and encapsulate it.

If you mean things that formally create a data race, then C++ cannot do that either.

[–] tcfhgj 4 days ago

Why can Rust not do that?

[–] gmueckl 4 days ago

How do you create shared mutable ownership of overlapping ranges of a single flat array in rust?

[–] tcfhgj 4 days ago

similarly except the algorithm to Mutex<[T]> and Arc?