back
77 comments
Of course since the old syntax is merely deprecated and not removed, going forward you now have to know the old, bad form and the new, good form in order to read code. Backwards compatibility is a strength but also a one-way complexity ratchet.

At least they managed to kill `auto_ptr`.

I doubt it will be a problem in practice.

Regular variadic arguments in general aren't used very often in C++ with exception of printf like functions. Not rare enough for majority of C++ programmers to not know about them, but definitely much more rare than their use in python. Main reason people know about it at all is printf. The "new" C compatible form has been supported since the first ISO standardized version of c++ if not longer. There haven't been a good reason to use the "old" form for a very long time. Which means that the amount of C++ code using deprecated form is very low.

Being deprecated means that most compilers and linters will likely add a warning/code fix suggestion. So any maintained project which was accidentally using C incompatible form will quickly fix it. No good reason not to.

As for the projects which for some reason are targeting ancient pre ISO standard c++ version they wouldn't have upgraded to newer standard anyway. So if new standard removed old form completely it wouldn't have helped with those projects.

So no you don't need to know the old form to read C++ code. And in the very unlikely case you encounter it, the way for accessing variadic arguments is the same for both forms through special va_list/va_arg calls. So if you only know the "new" form you should have a pretty good idea of whats going on there. You might lookup in references what's the deal with missing coma, but other than that it shouldn't be a major problem for reading code. This is hardly going to be the biggest obstacle when dealing with code bases that old.

The “new” form has been valid since the original 1998 C++ standard, where it was added for compatibility with C. “You now have to know” has therefore already been the case for the past 27 years. Back then the old pre-standard form was kept for backwards compatibility, and is only now being deprecated.
The old-style variadics are rarely seen in C++ these days, never mind this particular edge case. If you working in a vaguely modern version of C++ this largely won’t impact you. You can almost certainly ignore this and you’ll be fine.

Unless you have a massive legacy code base that is never updated, C++ has become much simpler over time. At a lot companies we made a point of slowly re-factoring old code to a more recent C++ standard (often a couple versions behind the bleeding edge) and it always made the code base smaller, safer, and more maintainable. It wasn’t much work to do this either.

To some extent with C++, complexity is a choice.

PyCuda 2024, used fairly often in certain industries, still contains `auto_ptr` ;-;
I think Rust has shown a way to remove deprecated interfaces while retaining back compat - automated tooling to migrate to the next version and give a few versions for a deprecated interfaces to stick around at the source level.
If you're talking about editions, this isn't how they work at all; every edition continues to be supported forever. (The part about automated migration tooling is true, and nice.)

There've been a few cases where code was unsound and should never have compiled, but did due to compiler bugs, and then they fixed the bugs and the code stopped compiling. These were handled through deprecation warnings with timelines at least several months long (Rust releases a new version every six weeks), but usually didn't have automated migration tooling, and didn't fracture the language mostly because they were rare edge cases that most programmers didn't encounter.

Editions are still allowed to remove old syntax or even remove APIs - they only can’t break ABIs. So the code is still there once removed from an edition in previous editions, but such symbols don’t even get linked if they’re unused supporting progressive removal. And similarly, I could see editions getting completely removed in the future at some point. Eg rather than indefinitely maintaining editions, in 20 years have a checkpoint version of a compiler that supports the previous 20 years of editions and going forward editions older than 10 aren’t in the build (for example, assuming a meaningful maintenance burden, which is hard to predict when that happens and what a specific policy looks like).
C++ almost never removes features because of the ABI compatibility guarantees. Programs compiled with older versions of the standard can be linked against newer versions.

This is allegedly because in the 80s companies would write software, fire the programmers, and throw the source code away once it compiled.

Fixing syntax by definition does not affect the ABI. And Rust has shown that both ABI and API compatibility can be achieved in the presence of several "versions" (editions) of the language in the same build.
Rust is a single vendor. It's not really the same situation.
Having multiple compiler vendors is a problem IMO not a feature. It fragments the ecosystem - the code compiles fine with this compiler but not this other one. The maintenance of portable Rust code is significantly easier.

I think the way forward is multiple backends (LLVM + GCC) to improve platform support, but a single unified frontend that works correctly on all platforms is a good thing.

This seems pretty good to me just on the level of trying to read C as someone using C++. Parameter packs and variadic templates are easily the most confusing syntax in C++ and cleaning it up is... very welcome
Hats off for using "..." in your comment immediately after a valid identifier word and with no comma in between, given the topic of the article.
Hats off for noticing. Not to be taken for granted in an increasingly skimming-oriented world
I used to slay with this in code golfing competitions from TopCoder, where you had to implement a function to solve a particular problem, thanks to C pointer maths and the gcc generally putting function arguments in order in the stack.

Turns out, these two are equivalent in practice (but UB in the C++ standard):

    double solve(double a, double b, double c, double d) {
      return a + b + c + d;
    }

    double solve(double a ...) {
      return a + 1[&a] + 2[&a] + 3[&a];
    }
> Turns out, these two are equivalent in practice

Not in the x86-64 SysV ABI they aren’t. The arguments will be passed in registers (yes, even the variadic ones), so how your compiler will interpret 1[&a] is anybody’s guess. (For me, x86_64-unknown-linux-gnu-g++ -O2 yields, essentially, return a+a+a+a; which is certainly an interpretation. I’m also getting strange results from i686-unknown-linux-gnu-g++ -O2, but my x87 assembly is rusty enough that I don’t really get what’s going on there.)

Example compiler explorer view: https://godbolt.org/z/b5z3q1616

Clang does the sensible thing with UB and just returns poison (a form of undefined value) in both cases, which manifests as do nothing on x86-64 and load a zero value on i386, because you need to push something on the stack and fldz is one of the cheapest ways to push something. Meanwhile, gcc is in both cases for the UB variant returning a + a + a + a;

FWIW, going back through older gcc versions, it seems i386 gcc stops implementing 'add the arguments' in version 11.1, although it's not until 15.1 that it has a sensible assembly for 'a + a + a + a'. The x86-64 gcc version is broken in 4.0 (where it stops copying the register arguments to the stack when va_start isn't called, I guess). Then it's adding xmm0 to the top 3 values on the stack until 11.1, when it's adding 'a + a + a + a', although not sensibly until version 15.1.

> return a+a+a+a; which is certainly an interpretation.

Zero is the only valid index of &a, so I presume the compiler just assumes that all the indexes in 1[&a] + 2[&a] etc must be zero. Even though they're in this case compile-time constants – the optimizer could check but why bother given that it's UB anyway. I assume modern C/C++ compilers have some flag to diagnose indexing that's known to be OOB at compile time.

K&R syntax is -1 char, if you are in C:

    double solve(double a,double b,double c,double d){return a+b+c+d;}
    double solve(double a...){return a+1[&a]+2[&a]+3[&a];}
    double solve(a,b,c,d)double a,c,b,d;{return a+b+c+d;}
>turns out these two are equivalent in practice

Err no; https://gcc.godbolt.org/z/sW3ea58oc

They're equivalent on GCC.

    double solve(double a[]) {
      return 0[a] + 1[a] + 2[a] + 3[a];
    }

    solve((double[]){1, 2, 3, 4});
The cast in the invocation can be macro-ed away. And the best thing is, the actual stack layout and data movement/shuffling is pretty much identical to the approach with <stdargs.h>, and with no UB or compiler intrinsics.
That's a compound literal, not a cast.
I'm far from C++ but reading this article confused be, from the form to the impact to the dead link https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p12...

I guess that's a preview how C++ require a lifelong commitment.

That should probably link https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p12...

The difference being that P1219R2 was actually a revised proposal from 2019 not 2021.

Yes, but no. I learned C++ in '90s when it was C with classes and some other noise added by Stroustrup. During the some 25 years that followed it had became a mess that's insanely hard to work with. I'm not going back to this language. I prefer plain C or Rust, leaning towards Rust when I fully comprehend the lifetime and borrow checker. Or when I have the luxury of having a GCed runtime, then the .NET with its easiest C# language with wonderful abundance of great libraries is the best choice. Nobody was ever fired for using .NET (for right purposes).
Tiring how often this needs to be said, but if you want "C with classes", you can just use C++ that way.

I've been using C++ for more than 30 years (I added thread_local to Cfront back in the early 90s), and while the language has grown dramatically in that time, there is fundamentally nothing that would prevent me from writing "C with classes" using the modern version.

I don't do that because I also like RAII, and polymorphism, and operator overloading and ...

I've never used .NET and could not imagine any scenario under which I would. The libraries that matter to me are mostly written in C or C++ and there are more of them than I'd ever need, mostly.

> nothing that would prevent me from writing "C with classes" using the modern version.

Or indeed "C without classes", just with some extra type-checking.

And of course K&R used Stroustrup's C++ compiler to build and test the code for TCPL 2nd Ed.

Languages are both read and written, restrictions like OP is pining for are fundamentally for reading. As such, it is not terribly helpful that they can opt in to restrictions when writing.
This is a good change. Potentially ambiguous to read syntax is being made clearer in a way that harms no previous code substantially.

I sometimes wonder if the comments that say nothing more than "C++ is too complicated" are from people who use it regularly, much less people who are even commenting on TFA

I don't fully understand the connection between your two paragraphs. It's not inconsistent for the language to be too complicated for it to be a good change.

I'm also not totally convinced that someone not using a language regularly means their view on it being too complicated must be invalid; I would fully expect someone who views it as too complicated to try to avoid using it, and there are enough people doing that, it might be a cause for concern. That doesn't necessarily mean they're right, but without further context it also doesn't mean their opinion isn't relevant.

Clarity and Elegance of Syntax > Backwards compatibility.

In my opinion anyway. C++ feels so bloated these days.

C++ seems to be constantly getting complicated. If the major version were to change, there wouldn't be any need for backward compatibility with the existing code, and it would have been okay to delete that syntax while creating an automatic formatter.
>If the major version were to change, there wouldn't be any need for backward compatibility with the existing code,

I have no idea where you get this idea from. I expect gcc v28 to be able to compile C++ from 2008, and I'm not alone in that.

I completely agree with you.

That said, I wish code written for gcc v28 didn't have to be binary compatible with C++ that was last compiled in 2008...

Are you suggesting we move to C++++?
It's already there. It's called C#
Personally I like C+. Picking the nice parts of C++, but skipping all the nonsense. I just wish C++ hadn't deliberately screwed up designated initializers with mandatory ordering. The C version of it that allows out-of-order assignments is clearly superior.
Didn't you read the article? That form is deprecated. The recommended one is C,++,++
I mean this is extremely minor, but the more incompatibility you create the slower the uptake will be. In the extreme you could create a Python 2->3 situation over old syntax.
Note to the author of TFA: The moment I get a sign-up modal on any site, I close the tab and move on.
C++ got too complicated after C++23 I went back to C.
You can always restrict yourself to a subset of C++ that takes advantage of RAII (resource handling is extremely painful in C), and get performance benefits like move semantics, without the insanely complex stuff.

I love C, but C++ has worthwhile advantages even if you heavily restrict which features you use.

I do this and my C++ is really like python to code in. But, if you work in a team, there will always be a bunch of people who come in and start using all sorts of exotic shit for their resume or hobby interest.
No offense intended to your perspective, but I do find it a little amusing that C++23, which was generally considered a disappointingly small update due to COVID complications, was the breaking point in complexity.
C is the new vinyl.
learned something new. thanks for the article.