C++ will always stay relevant. Software has eaten the world. That transition is almost complete now. The languages that were around when it happened will stay deeply embedded in our fundamental tech stacks for another couple decades at least, if not centuries. And C and C++ are the lion's share of that.

COBOL sticks around 66 years after its first release. Fortran is 68 years old and is still enormously relevant. Much, much more software was written in newer languages and has become so complex that replacements have become practically impossible (Fuchsia hasn't replaces Linux in Google products, wayland isn't ready to replace X11 etc)

▲

pornel 5 days ago | parent | next [-]

It seems likely that C++ will end up in a similar place as COBOL or Fortran, but I don't see that as a good future for a language.

These languages are not among the top contenders for new projects. They're a legacy problem, and are kept alive only by a slowly shrinking number of projects. It may take a while to literally drop to zero, but it's a path of exponential decay towards extinction.

C++ has strong arguments for sticking around as a legacy language for several too-big-to-rewrite C++ projects, but it's becoming less and less attractive for starting new projects.

C++ needs a better selling point than being a language that some old projects are stuck with. Without growth from new projects, it's only a matter of time until it's going to be eclipsed by other languages and relegated to shrinking niches.

	▲	pjmlp 4 days ago \| parent [-]
		It will take generations to fully bootstrap compiler toolchains, language runtimes, and operating systems that depend on either C or C++. Also depending on how AI assisted tooling evolves, I think it is not only C and C++ that will become a niche. I already see this happening with the amount of low-code/no-code augmented with AI workflows, that are currently trending on SaaS products.

▲

ktpsns 5 days ago | parent | prev | next [-]

As long as people write software (no pun intended), software will follow trends. For instance, in many scientific ecosystems, Matlab was successfully replaced by Scipy. Which happens to get replaced by Julia. Things don't neccessarily have to stay the same. Interestingly, such a generational trend currently happens with Rust, despite there has been numerous other popular languages such as D or Zig which didn't have the same traction.

Sure, there are still Fortran codes. But I can hardly imagine that Fortran still plays a big role in another 68 years from now on.

▲

klik99 5 days ago | parent | next [-]

Matlab/Scipy/Julia are totally different since those function more like user interfaces, they are directly user facing. You're not building an app with matlab (though you might be with scipy and julia, it's not the primary use case), you're working with data. C++ on the other hand underpins a lot of key infrastructure.

▲

gmueckl 5 days ago | parent | prev | next [-]

I am not saying that these languages will stay around forever, mind you. But we have solidified the tech stacks involving these languages by making them ridiculously complex. Replacement of a programming language in one of the core components can only come through gradual and glacially slow evolution at this point. "Rewrite it in XYZ" as a clean slate approach on a big scale is simply a pipe dream.

Re Matlab: I still see it thriving in the industry, for better or worse. Many engineers just seem to love it. I haven't seen many users of Julia yet. Where do you see those? I think that Julia deserves a fair chance, but it just doesn't have a presence in the fields I work in.

	▲	ants_everywhere 5 days ago \| parent [-]
		I've heard via former employees that Mathworks has conceded that Python ate Matlab's niche and that they're focusing on Simulink

▲

binary132 5 days ago | parent | prev | next [-]

You’re thinking of software that is being written today. GP is talking about software we use every day in every device on the planet that hasn’t changed since it was written 30+ years ago.

	▲	nine_k 5 days ago \| parent [-]
		What is this software? E.g. Linux is 33 years old; barely a few percent of Linux 1.0 remains in a modern kernel, if we count lines of code. Maybe GNU Emacs has a larger percentage remaining intact; at least it retains some architectural idiosyncrasies from 1980s. As of Fortran, modern Fortran is a pretty nice and rich language, very unlike the Fortran-77 I wrote at high school.

▲

wolvesechoes 5 days ago | parent | prev | next [-]

> For instance, in many scientific ecosystems, Matlab was successfully replaced by Scipy. Which happens to get replaced by Julia

If by scientific ecosystems you mean people making prototypes for papers, then yes. But in commercial, industrial setting there is still no alternative for many of Matlab toolboxes, and as for Julia, as cool as it is, you need to be careful to distinguish between real usage and vetted marketing materials created by JuliaSim.

▲

vhcr 5 days ago | parent | prev [-]

Scipy is a wrapper of Numpy, which is a wrapper of C and Fortran.

▲

briHass 5 days ago | parent | prev | next [-]

Especially the 'backend' languages that do all the heavy lifting for domain-specific software. Just in my vertical of choice, financial software, there are literally billions of lines of Java and .NET code powering critical systems. The code is the documentation, and there's little appetite to rewrite all that at enormous cost and risk.

Perhaps AI will get reliable enough to pour through these double-digit million LOC codebases and convert them flawlessly, but that looks like it's decades off at this point.

▲

casey2 5 days ago | parent | prev | next [-]

I'm not so sure. The user experience has really crystallized over the years. It's not hard to imagine a smart tv or something like it just reimplementing that experience in hardware in the not too distant future (say 2055 if transistor and memory scaling stall in 2035).

We live in a special time when general processing efficiency has always been increasing. The future is full of domain specific hardware (enabling the continued use of COBOL code written for slower mainframes). Maybe this will be a half measure like cuda or your c++ will just be a thin wrapper around a makeYoutube() ASIC

Of course if there is a breakthrough in general purpose computing or a new killer app it will wipe out all those products which is why they don't just do it now

▲

worik 5 days ago | parent | prev [-]

> Software has eaten the world.

Bit off more than it could chew, no we all have indigestion