The beauty and simplicity of the good old C-style void* in C++

https://giodicanio.com/2026/06/05/how-to-declare-a-c-plus-plus-function-that-takes-a-blob-of-memory/

by movd128 • 2 months ago

68 174 2 months ago

gignico2• 1 month ago

It seems that some people are really losing the taste for good readable code.

It seems that some people never had taste for good reliable code. Use `void ` and now any error whatsoever is a direct undefined behavior. Moreover `std::span` clearly says that you are not* taking ownership of the memory (even though the language does not check it of course), while `void *` does not.

I understand that people can have many things to say about C++, and I do as well, but `std::span` should have been there decades ago and is such a life saver in these situations. A truly zero-cost abstraction which effectively saves you from a lot of troubles.

spacechild17• 1 month ago

but `std::span` should have been there decades ago

Absolutely! I now use it consistently in all new projects where I can afford to mandate C++20. I guess nobody bothered to make a proposal before...

pjmlp16• 1 month ago

They did in C, from one of the language authors even, and it was not accepted.

https://www.nokia.com/bell-labs/about/dennis-m-ritchie/varar...

By the way, both Extended Pascal, Mesa/Cedar and Modula-2 have them, under the name of open arrays.

Basically it took Go, C# and others for C++ to finally get its span.

C probably never will.

trumpdong4• 1 month ago

There's lots of UB in C-family execution models. Some of which is not actually UB because the implementation defines it - e.g. aligned DWORD-sized memory access is atomic on Windows because Microsoft said it is.

By choosing to use this language you choose to navigate the UB. Otherwise you'd be writing in Go, or Python.

It is possible to write reliable code despite the presence of UB in a language just like it's possible to drive to work every day for 20 years despite most of the directions you can point the car leading to an immediate crash. That's a needle with a much thinner eye than UB in C, and most people manage it. Mainly it means being very careful about lifetime and ownership. The Linux kernel manages it 99% of the time simply by being careful about lifetime and ownership, and that's a project with a huge number of contributors who don't intimately know each other's modules. I'm the Linux kernel you can't just say "new whatever" - you must have a plan for a lifetime of that whatever, and other people will review it.

I agree with you about std::span.

arcticbull2• 1 month ago

Yeah but also, quick question:

  struct S {
      char c;
      int i;
  };

  struct S a = {0};
  struct S b = {0};

  memcmp(&a, &b, sizeof(a)) == ...

If you answered 0, you'd be wrong, the answer is undefined, thanks to padding, initialization and alignment rules. Padding bytes are undefined, and not guaranteed to be initialized to zero even if the variable is declared static (where the members would be zeroed).

This is why the compiler is angry at the post writer, and why the reinterpret_cast is needed. Ideally if they wanted to do something with the data, they'd unbox the structure.

That's why it's not a good idea to use void* to pass arbitrary data interchangeable with bytes. It's a location, it makes no representation as to what's there and how to interact with it. Let alone who owns it.

std::span solves two problems here. One is the ownership problem. The other is that span<T> is a T[]. void* is god only knows.

The post asserts:

The code is very clear and straightforward: you pass a pointer to the custom data structure, and its size in bytes. That’s it. Simple and clear.

This is unfortunately entirely false in C thanks to the aforementioned alignment/padding UB (and of course inner pointers). This is addressed with std::span. You'd still have to reinterpret_cast your structure to get the UB.

Why should people complexify and uglify their C++ code with the uint8_t pointer (or std::byte), when void* works just fine??

tl;dr: because it doesn't. It just kinda looks like it does if you squint, and it's going to lead to the gnarliest bugs in the world.

repelsteeltje4• 1 month ago

There is a difference between UB in C, and something being undefined in some version of Microsoft C on Windows.

Many of C's UB is specifically, intentionally left undefined in the standard to express code that relies on some specific way it is handled, is not proper, portable C. Indeed, the DWORD-sized memory access being atomic doesn't apply to MS Windows prior to version 3.0 running on a 80286.

It's UB because the ISO C spec says it's UB.

pjc502• 1 month ago

Some of which is not actually UB because the implementation defines it

No - if something is UB in the spec, it's UB. The implementation will do something, sure, but what it does is not fixed and may even change based on compiler version and optimization level.

DWORD-sized memory access is atomic on Windows because Microsoft said it is

Well, Intel said it is. Mind you I don't think there are any 32-bit native architectures where aligned dword access isn't atomic. Unaligned, on the other hand ...

locknitpicker8• 1 month ago

I understand that people can have many things to say about C++, and I do as well, but `std::span` should have been there decades ago (...)

Decades is kind of a stretch. C++11 introduced smart pointers, and finally getting C++0x out of the door was already a major victory. Given the history of C++, it would be unrealistic to introduce something like std::span before C++17.

Meantime, some organizations are still struggling to migrate to something like C++14.

pjmlp16• 1 month ago

It could have been there since the beginning, given that open arrays (aka spans) already existed in other languages, and there was even a failed proposal from Denis Ritchie regarding C.

The C++ span proposal came from Microsoft,

https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p01...

gignico2• 1 month ago

Afaik std::span does not need anything that was not in C++98 already, or am I missing something?

delta_p_delta_x3• 1 month ago

A truly zero-cost abstraction

Sadly the MSVC ABI makes std::span and std::string_view a pessimisation:

https://github.com/tringi/win64_abi_call_overhead_benchmark

https://godbolt.org/z/7baaox7re

usrnm2• 1 month ago

Sounds like a compiler bug to me. It is a valid reason to avoid them in some rare cases right now, but it doesn't make the feature itself bad

pjmlp16• 1 month ago

That is quite common among C developer culture, play loose and brace for impact.

voidUpdate• 1 month ago

"An interesting question you may ask in C++ is: “How would you declare a function that takes a blob of memory as input?”"
"Now, suppose that you want to pass to this function a custom structure, like this:"

You would create another function that actually works based off that structure, rather than using your first function which operates on a set of bytes in memory. That way it's readable, like they want, and type-safe

trumpdong4• 1 month ago

I find this to be a snarky non-answer. You really think everyone should write their own memcpy for every POD type they want to memcpy?

mfost• 1 month ago

There's no need: there's std::copy already.

Or maybe the idea was to create a typesafe template wrapper around the generic function which is also very common and really nice. No need to create one wrapper per type, a single template should work.

AnimalMuppet4• 1 month ago

In addition to what mfrost said, there's also no need because C++ assignment is member-by-member copy unless otherwise specifically implemented for the type. If you have a POD, then that's what you get with assignment; there's no need to call memcpy at all.

(The difference is that memcpy will copy padding bytes, and the assignment operator may not. But if you depend on the values of the padding bytes, you have major problems...)

themafia• 1 month ago

I'm not a fan of C++ precisely because of template noise but what you gain with span, in that the pointer and the length are joined together, seem to outweigh the complaints on style.

Isn't there a way to make this an alias anyways?

arcadialeak• 1 month ago

char* is an exception to strict aliasing rules of C++ precisely to facilitate the author's use case. You would still need a reinterpret_cast to make it work, but it's actually good because it makes the intent clearer, and the cast would have still happened either way to read the raw bytes.

quietbritishjim• 1 month ago

That was my first instinct too, but nothing the author said indicates they actually need non-strict aliasing. If the function had been:

   void DoSomething(void* src, void* dst, size_t numBytes);

... then it would be a different matter since maybe you want to allow src and dst to alias. Although, even then, they're still allowed to alias so long as the function accesses them both through char*, so the function signature can still use void*.

(Going deeper, non-strict aliasing applies to any pointers of the same type passed to a function. So if src and dst were both cast to float* inside the function, and if they really are both of that type (technically "an object of type float exists at the pointed-to location) then they can still alias. The char* exception is the only case that you can access a memory location through two different types of pointer and they can still alias.)

It's interesting the author mentions uint8_t. It's certainly more explicit than char, but it doesn't have the same aliasing guarantee (very strictly speaking - in practice it's almost always an alias for unsigned char or char, which does).

myrmidon• 1 month ago

This is actually pretty annoying in embedded programming in C, because you'd often really prefer to use a uint8_t buffer[] for serialization functions (e.g. to write arbitrary data on some bus etc.) over char*, but you'd actually lose the aliasing permissiveness that you need (if you are strictly sticking to the standard-- this is often ignored in practice).

stinos2• 1 month ago

Why should people complexify and uglify their C++ code with the uint8_t pointer (or std::byte), when void* works just fine??

Fair point (although to be honest: 'complexify' feels a bit of an exaggeration here to me), but the answer to this why is simple: document and express intent clearly. The compiler gave you an error first such that you're forced to consider what you're doing. Any seasoned C++ developer seeing this knows what this reinterpret_cast means.

Wow. With std::span the complexity-meter bumps in the red zone and goes even higher!

Same remark: yes, it's a bit more text to read, but again: to me (and many others I'm guessing) this clearly expresses intent. I also do not find it particularly hard to read. I mean, it's C++, you're likely going to encounter templates at one point or another, except in super specific software perhaps. But no-one also ever argued the C++ learning curve was easy, and trying to make it easier by refusing to use features which were added for good reasons and instead going back to constructs which are the very source of those reasons seems a bit backwards.

As a nice addition, if you use SAL annotations, the function could be decorated a bit to help code analyzers detecting memory bugs

Some might also say it complexifies and uglifies the code. And in any case makes it non-portable on top of that.

repelsteeltje4• 1 month ago

And SAL annotations aren't even C++ proper.

VulgarExigency3• 1 month ago

It seems unlikely that this is the case, as the author appears to be experienced, but the post reads like the author has never had to maintain a "simple" and "beautiful" function that was mangled into incomprehensibility over the years, and where if a more expressive type signature had been written from the start, it would have restricted the damage caused over time.

zahlman6• 1 month ago

a "simple" and "beautiful" function that was mangled into incomprehensibility over the years, and where if a more expressive type signature had been written from the start, it would have restricted the damage caused over time.

...Can you give a concrete example? I've been programming literally since the 80s and that doesn't ring true at all for me.

thomasmg• 1 month ago

I don't have a strong opinion what is better in this case, but my view is:

document and express intent clearly

Arguably, the void* does that as well?

Any seasoned C++ developer seeing this knows what this reinterpret_cast means.

Same for void*?

it's a bit more text to read

If you have to call it many times, this adds up.

Some might also say it complexifies and uglifies the code

I think the point is that it adds security, which the other options don't. And, it doesn't add complexity on the caller, but only at one place: the implementation.

makes it non-portable on top of that.

This can be solved.

stinos2• 1 month ago

Arguably, the void* does that as well?

Sort of (I mean: seeing void* and a size probably means 'arbitrary sequence of bytes' or something like that, but well it's void* so it can be like anything whereas with std::span you get more of a hint what's going on just based on the type), but not at the callsite which is what the author is referring to when it's about reinterpret_cast.

I think the point is that it adds security, which the other options don't

Imo span also does that to some extent, but already when writing the code and not afterwards in e.g. static analysis. E.g. if I get an std:span<const char> I'd have to do counterintuitive things to misuse it. Annotating a void* still leaves it a void* which I then need to cast to char* if I think that's what it is intended.

Don't get me wrong: I've written my fair share of void* but these days I really feel like there's almost always a better thing which can be used instead. Though I do admit that since I've written and consumed a lot of code with such alternatives I'm not hindered by readability/apparent complexity of it anymore but I understand that's not the same for everyone.

throwaway274485• 1 month ago

Arguably, the void* does that as well?

How do you figure? The type is a pointer to quite literally anything, including nothing (ie a pointer that cannot be dereferenced). If you're working with bytes, indicate this with the type.

drivebyhooting6• 1 month ago

Span will increase compilation time for no useful reason. It’s not any safer at the call site.

throwaway274485• 1 month ago

Well if compilation time is an issue, you chose the worst possible language to use. But if you must use C++, you should use the mechanisms that best communicate intent.

ndiddy2• 1 month ago

Span lets you use a ranged for loop to iterate over the contents without worrying about exceeding the bounds, which is safer than pointer+size if that's all you'll be doing. C++26 also introduces .at() for span, and the new hardened standard library enforces bounds checking when using operator [] on a span.

MaulingMonkey4• 1 month ago

Fair point (although to be honest: 'complexify' feels a bit of an exaggeration here to me)

Both uint8_t and std::byte require a header (<cstdint> or <cstddef>) which may expose you to platform x config specific build failures if you do any conditional #including, and the latter is a whole damn enum class with a strange adversion to arithmetic, where `byte |= 1` becomes `byte |= std::byte(1)`, `byte += 1` becomes `byte = std::byte(std::to_integer<std::uint8_t>(byte) + 1);`, and both become something you can accidentally step into in your full debug builds because it's an actual function call (at least on MSVC - still extra instructions on clang/gcc, but I can see the dang call instruction on MSVC!) instead of a compiler built in.

Not to mention, neither is vanilla C++03... I threw a `std::byte` example in a quick godbolt snippet and MSVC wouldn't compile without adding /std:c++17, because of course it defaults to earlier. Which is silly, but that's also the story of my life.

And don't get me wrong - that's all relatively minor - but it's all for middling to negative value IME. `void*` is frequently clearer - it's a signal that it's an opaque blob at this point in the code, and that something else will try to give it meaning later. I struggle to think of a single bug that I've encountered, that would've been caught by the compiler had I used `std::byte` over `unsigned char` or `void`. And conversely, I've seen APIs accepting `std::byte` but requiring higher alignment, where with `void` I might not have dropped my guard as much.

`std::span`

At least manages to bind pointer and size into a single variable, which IME at least has the advantage of eliminating some bugs (e.g. mismatching pointers and sizes) and allowing some nifty utility functions to become a lot more wieldy. You can do things like feed it an array and not have to do any of your own `sizeof(...)` shenannigans. At this point you're possibly getting into positive expected value, but I'm going to eye roll at pull requests refactoring `void*` based stuff to use it unless I see at least one actual concrete example of calling code improving alongside it - I don't want just hypothetical theoretical ergonomics, I want actual concrete ergonomics!

AnimalMuppet4• 1 month ago

`void*` is frequently clearer - it's a signal that it's an opaque blob at this point in the code, and that something else will try to give it meaning later.

And that's fine, until something else gives it the wrong meaning later. If you're just plumbing, and you're pumping around opaque blobs, if somewhere in the plumbing you connect the wrong source to the wrong destination, you get no warning.

account422• 1 month ago

and the latter is a whole damn enum class with a strange adversion to arithmetic, where `byte |= 1` becomes `byte |= std::byte(1)`, `byte += 1` becomes `byte = std::byte(std::to_integer<std::uint8_t>(byte) + 1);`, and both become something you can accidentally step into in your full debug builds because it's an actual function call (at least on MSVC - still extra instructions on clang/gcc, but I can see the dang call instruction on MSVC!) instead of a compiler built in.

That's the point, std::byte is for opaque bytes. You're not expected to do arithmetic directly just like you can't do arithmetic on void.

arka2147483647• 1 month ago

The best part of void* is that it is very terse. Both in definitions, and in access.

All cpp alternatives are more wordy.

I wonder how this conversation wound go if the was an as terse, but also typesafe cpp alternative.

squirrellous• 1 month ago

One could argue the reinterpret_cast makes the intent more explicit which is a good thing.

That said I don’t have much against the use of void* or even char* here. If it works in C, it works in C++ just fine. std::span is not the right tool for this.

delegate• 1 month ago

It depends on what your function does with that memory. If the fn expects any kind of structure at that address, you and your callers are on your own, compiler can't help if the caller passes the wrong thing. Worse, acessing that memory might not immediately crash, but lead to strange side effects in your program.

Dynamic languages can handle this with reflection, but with void* you can only pray nobody makes the mistake..

delta_p_delta_x3• 1 month ago

The blogger and the blog says:

BTW: As a nice addition, if you use SAL annotations
Windows C++ Programming

Not everyone will see the irony, but the Windows user-mode application and library suite and the kernel now very heavily rely on the safety mechanisms of C++ that the author calls 'complex', 'uglif[ied]', and has 'los[t] the taste for good readable code'. I'm of course referring to the Windows Implementation Library: https://github.com/microsoft/wil This is explicitly an effort from MS WinDev to make Windows C++ code safer. User-mode applications writing native Windows code can and absolutely should use it, too.

Any time I see `void*` in C++ I ring-fence it as a C-ism and make sure I `reinterpret_cast`. For me, a bag of bytes is `std::span<std::byte>`. void* is a memory location with no provenance, no ownership, no size information, nothing. Do I even know if it is this program's memory, or some shared memory construct, or maybe even a pointer into GPU memory? No for all.

C likes to play fast and loose and its proponents call it 'beautiful and simple', I call it a segfault/use-after-free/double-free waiting to happen.

pjmlp16• 1 month ago

It goes even further their beloved C code is compiled in compilers written in C++, including the standard library, exposing C++ implementations as extern "C" functions.

It is a pity that Microsoft backtracked on their C support.

WWDC is happening this week, one set of announcements at State of the Union was how Apple replaced a few C, Objective-C and C++ components, including at OS level with Swift.

repelsteeltje4• 1 month ago

Interesting. I'm know nothing about Apple, but maybe you can explain how idiomatic Swift handles Blobs and how that interfaces with C or C++ around void ptrs, std::spans etc.?

_the_inflator• 1 month ago

I think that the author is right in everything he says and yes, there is beauty in it.

However, the antithesis is also correct that there exist better solutions to solve the issues.

Both premises hold true.

I have an extensive assembler coding background on 6510, M68000, and i486. I had a very hard time accepting that something could be solved faster and more stable in a higher order language while the downside is more memory, more CPU etc.

More and more it turns out that programming languages are something accidentally read by machines and written by humans, even though this premise got destroyed lately by AI.

However, what I love about C++ is, that it has a basic canon of commands that can be used to build nearly everything while looking extremely ugly and hard to grasp if you don't read very slowly and accurately - so it is a very error prone and dangerous thing that rightfully got substituted by better constructs that allow for better distinctions as well as usage.

I could do everything in assembler (Hey Python users: you know that in the end everything ends up as machine code, don't you?) but it takes 100x times longer and is constantly reinventing the wheel.

Have you ever started to get into the intricacies of bit signs? No? Well, you should definitely, and to this day it gave me a lasting impression when I started wrapping my head around it, when I was 10 to 11 years old hacking my way into the world of assembler programming on C128.

You don't want to take every concept into consideration. You don't want to take interoperability into consideration. All the time!

You want to focus on the problem to solve, not the implications of the implementations all the time.

I am having such a blast very often using Python since it just works with much cognitive distraction about which language construct to use in order to get the machine doing what you want. It is so capable, enable it, to simply ensure within boundaries that the compiler uses the best decision given the context, which is up to analysis.

That's why I stopped using C++ or more precisely stopped any attempts and trying to be smart or fancy. I got to re-read and maintain the code month to years later and history showed, I don't marvel at how magic the line works and brutally smart I was at the time, but simply hate me for obscuring something in a line, that could be well understood if I had used 10 lines, while the compiler gives a damn anyway.

C++ is still necessary but every discussion to this day is about the point you made: every digit counts - and also which position, context etc. You got to be very prolific in order to put into a line what other put into 10.

Is it worth it? No.

In early days it was the correct decision. Memory was sparse, CPU power slow, and the language was small compared to today.

The last time I felt comfortable with a "assembler kind feeling" was with JavaScript before ES6. Peak jQuery level, with the most coolest concept only JavaScript has: Function.prototype.toString()

John Resig will have his place in my programming heroes olymp, who revealed this secret for me, and it opened my eyes for the beauty of higher order languages.

I admire C++, but so do I Python.

But I hope I won't have to ever use C++ again.

pjc502• 1 month ago

One of these days I want to do a "typesafe macro assembler" that actually is the language people think that C is.

bcjdjsndon4• 1 month ago

You could call it rust and just rename the remaining UB as "unsafe" and call the problem solved

trumpdong4• 1 month ago

You could even give it identical syntax to C!

pjmlp16• 1 month ago

It already exists, MASM and TASM, and related derivatives, with high level control flow pseudo instructions and less UB.

rfgplk2• 1 month ago

I'm actually working on something similar to this. Basically works as an additional preprocessed layer over C++. You can get some crazy results, but it's tricky to integrate with existing build tools without causing havoc.

repelsteeltje4• 1 month ago

There will always be cases (like audio processing, car brakes, pace makers) where hard real-time constraints prohibit GC languages (as well as l1 cache, instruction reordering and other optimizations). Also, consider that Python's performance frequently originates in it's bindings to libraries written in C, C++, Fortran, Rust.

I recently ran a few Java benchmarks and found that an array holding a bunch of objects incurred approx 3x the number of bytes compared to the expected number based on underlying class data structure. With current RAM prices, that is something to consider if you're building software that's meant to scale. Mileage may vary, but I expect JavaScript or Python will be similar.

So, sure. There is a case to be made that ergonomics and dev velocity. And premature micro optimizations might take your focus away from good systems architecture. But I've frequently found the need to peal of leaky abstractions and having to understand and be savvy at low level stuff too. Nothing wrong with studying the guts of a C64 or Amiga, today.

Python, Java or TypeScript are good educational tools, but you'd be doing yourself a disservice if you'd confine yourself to them without understanding computers.

simiones3• 1 month ago

Note that, while complex, there exist GCs that can handle both soft real time and even hard real time constraints - especially for Java. Memory overhead is a problem with GC languages, though, and that one is by design.

bcjdjsndon4• 1 month ago

What's your point here? The higher the language level, the less performant they are? Why are using anything but assembler then?

zahlman6• 1 month ago

There will always be cases (like audio processing, car brakes, pace makers) where hard real-time constraints prohibit GC languages (as well as l1 cache, instruction reordering and other optimizations). Also, consider that Python's performance frequently originates in it's bindings to libraries written in C, C++, Fortran, Rust.

Sure. And every Python programmer who has any interest in those use cases learns about the issues quickly. Or more to the point: a big chunk of them are things you'd only do if you were employed to do them, and employers are setting the language requirements already. And Python programmers in particular are well aware of compiled-language bindings; that's the reason they're trying to use the packages that make package installation non-trivial.

Huge swaths of use cases don't require performance.

Python, Java or TypeScript are good educational tools, but you'd be doing yourself a disservice if you'd confine yourself to them without understanding computers.

This is an extremely strange thing to say when replying to someone who just described having extensive experience with C++ and multiple flavours of assembly.

I recently ran a few Java benchmarks and found that an array holding a bunch of objects incurred approx 3x the number of bytes compared to the expected number based on underlying class data structure.

It was also strange to say that if you also had this experience yourself. A solid "understanding of computers" would have given you a better mental model of what Java needs to allocate. Results like this are because "the expected number" was not well thought out.

if you're building software that's meant to scale.

... And yet everyone just keeps pumping out Electron apps. Curious, that.

zahlman6• 1 month ago

Hey Python users: you know that in the end everything ends up as machine code, don't you?

I don't understand where you're trying to go with this call-out, especially if you're also describing yourself as a Python user.

But, like, no, not really; ordinarily, Python is bytecode-compiled and then the bytecodes are interpreted. There's machine code doing the interpretation, but that interpretation is not transformation.

adev_• 1 month ago

This post post is honestly speaking a bag of garbage and ill advises:

Some good old habit from C can still be positively used in C++, like the void* pointer and the size parameters.

That's garbage.

There is a clear interest of passing both size AND pointer in a single parameter like `std::span<std::byte>: It bind both value together and guarantee that you do not mess with the size of your buffer.

Pass "data" and "size" parameters through a chain of 5 function calls and there is a non-null probability that you passed "other_size" instead of "size" somewhere. This pattern happens everywhere in old C codebase and has been the source of countless security vulnerabilities and random buffer overflows for decades.

All modern languages (including freaking minimalist Golang) have now a "slice/span" concept built in.

It is not just to annoy programmers (and allow them to complain about 'complexity' in blog posts) but because it is a major improvement in term of memory safety and in term of reducing user errors.

It seems that some people are really losing the taste for good readable code.

If 'span<std::byte>' or 'span<char>' are unreadable for you. The problem is not span, the problem is you.

These are concepts that has been existing for decades in almost all modern programming languages.

Even in conservative C++, it exists since 2014 in the GSL, in Qt and in boost.

And the interface is no different from vector...no excuse here... It is itself the most basic data-structure in C++.

Why should people complexify and uglify their C++ code with the uint8_t pointer (or std::byte), when void* works just fine??

Sure. Let's extend the logic: I do propose also to replace all typed arguments with a void* pointer.

Because after all: 'It will just works fine' right ?

Type-safety and clear interface are overrated, we could all use only bytes and remove interface all together to get a closer experience of Fortran 77.

/irony

Or maybe something even more complicated, like this? template <typename T, std::size_t N> void DoSomething(std::span<T, N> data)

First that is non-sense.

If you want to pass a mutable buffer of byte, the correct signature is:

``void DoSomething(std::span<std::byte> data)``

There is no need for template signature here. You are making things up.

Second, there is also no need for the N parameter

``span<Type,N>`` is only used when enforcing a buffer with its size known at compile time is desirable. It can be for vectorization (e.g buffer is a multiple of the SIMD line) or to make it explicit in the interface (e.g for bloc cipher for instance)

states that the pointer points to input read-only memory (_In_reads_)

You do that by using `std::span<const std::byte>` in any C++ codebase.

The fact he brags about that as "an advantage" for separated parameter passing just show currently how little is known here.

My Pluralsight Courses

The kind of C++ code proposed in this blog post would be straight be refused in any PR in almost any serious organization with a proper review process.

So bragging about it on a blog while proposing some C++ teaching is audacious to say the least.

To finish on that.

The sad thing is that there would be very valid criticism on `std::span<std::byte>`:

- Span does not do boundary check on access by default. Which is a bad design decision in 2026.

- It has an impact on compilation time due to the header inclusion

- std::byte is annoying to work with because it is a hack around an enum instead of a proper C++ builtin type.

But the blog post misses all these points entirely and sticks to complaining about 'Old C being better' the same way your family Grand-Uncle still brags about 'lead gasoline being better' for his 70s Pontiac.

akkaygin• 1 month ago

In fact, std::span is a class template, and somebody would suggest to make the function that processes the generic memory blob a function template! Really? Something like this??

Yes.

api• 1 month ago

Real programmers use uintptr_t for pointers.

pwdisswordfishq• 1 month ago

“Hey, why do you use the unsafe old C-style void* pointer?

Exactly, one should avoid unnecessarily erasing pointer target types. Luckily, C++ gives much better tools for that than C. This should have been a tem—

Use some safe explicit type like uint8_t, which clearly represents an 8-bit byte!”

Sigh. Out of the frying pan, into the fire.

bcjdjsndon4• 1 month ago

Makes you wonder why OP is using cpp to begin with if theyre suggesting void*

bcjdjsndon4• 1 month ago

How in Nora does that get downvoted? It's the mainstream view on it

IX-103• 1 month ago

Ick. The entire article starts from the fundamentally flawed premise that "you want a function that takes a blob of memory as an argument". Then they discuss bytewise access into structures..

Passing around void pointers is simply not a safe thing to do in C++. You can't do anything with a void pointer, so you're probably going to cast it as something else. Use that type instead, so that your caller knows they need to pass a valid pointer to that type. If the pointer has the wrong alignment then that will result undefined behavior. If you need to support multiple pointer types, use templates.

And, unless there are some really weird circumstances, you actually don't want to access your structures bytewise. Offsets can shift with compiler flags/versions. If you want serialization , please use a serialization library that correctly handles all of the odd cases. These can be quite efficient.

I've only actually had to munge bytes in a class once. Somebody decided that a previously POD class that was passed between processors with different memory spaces needed a virtual function, so I had to overwrite the vtable when I received it to make it valid.

AnimalMuppet4• 1 month ago

In general, you're right.

The exception that I could think of is a "dump memory" function. You take a pointer to something (who cares what it is), and print out the bytes there. That I could see taking a void*.

But that's a really limited case. In general, yes, you do not want to be dealing with blobs of memory as arguments. You want to be dealing with things that are known to be the right kind of thing as arguments.

eddd-ddde• 1 month ago

But parent is right, you have to cast it anyways before reading from it, so might as well take the right type from the beginning.

zahlman6• 1 month ago

Many years ago I remember going through the Boost library and seeing C-style casts that seemed entirely gratuitous. I tried replacing them with what I was pretty sure were the equivalent C++ reinterpret_casts, and the result didn't compile. I never did figure it out.

s28l2• 1 month ago

C style cast can be either a static_cast or reinterpret_cast, but it can also be a const_cast or a static+const or reinterpret+const. Finally, it will perform a static_cast that bypasses private inheritance (because the alternative would be to fall back to a reinterpret_cast, which is wrong if the static_cast needs to apply an offset to the pointer)

zahlman6• 1 month ago

a static_cast that bypasses private inheritance (because the alternative would be to fall back to a reinterpret_cast, which is wrong if the static_cast needs to apply an offset to the pointer)

That may well have been it, then. I would think that if it could have been expressed naturally, it would have been.

(I don't think I ever used private inheritance in my own C++ code. I'm not a huge fan of inheritance at all, so.)

pjmlp16• 1 month ago

To add on top of sibling comment, C style casts are too lose, and the main reason for the new C++ style casts is improved type safety.

So instead of anything goes, there is some additional type checking depending on the type of cast being made.

mwkaufma2• 1 month ago

If you know you want bytes -- A void* of unknown provenance cast to anything other than char* is UB so just skip the middleman and use char*.

s28l2• 1 month ago

char* can also be a C-style string. std::byte has the same special treatment in the standard as char and unsigned char, with the added benefit of not being used for other purposes (i.e. ASCII character or uint8, respectively).

mwkaufma2• 1 month ago

I was trying to appeal to OPs maladaptive "C++ that looks like C is more legitimate" aesthetic preference, haha.

singpolyma32• 1 month ago

Next realize you can just use C instead of C++ at all!

cherryteastain• 1 month ago

In C++ you'd just do

    template <typename T>
    void DoSomething(T const& data);

or if T* is supposed to point to a tightly packed buffer

    template <typename T>
    void DoSomething(std::span<T> data);

as the author pointed out. I don't see how that is ugly or more complicated than the original void* approach.

There is no need to pass the size of T or length of the span, former is just a sizeof(T) away and latter is a data.size(); away.

In fact, a lot of codebases would outright ban the uint8_t* and reinterpret_cast trick the author is complaining about via clang-tidy rules.

jayd16• 1 month ago

Here's a thought experiment. Is void* something we should add to other languages?

Would anyone argue yes?

drysine• 1 month ago

a function that hashes some input data (using SHA-256, or whatever hash algorithm)

Along with padding bytes.

Why should people complexify and uglify their C++ code with the uint8_t pointer (or std::byte), when void* works just fine??

That was the intention of reinterpret_cast - make ugly code look ugly.

adrian_b2• 1 month ago

The "void*" of C solves a frequently encountered problem, but it has an inappropriate and misleading name, because such a pointer points to something, it does not point to nothing. Moreover, it does not have a size.

The correct solution in a programming language is to have a primitive bit string type (with a length that is a byte multiple) and to have a concise way (e.g. with dedicated symbolic operators) to write a type conversion from any data type to a bit string and a type conversion from a bit string to any data type.

Then the operations that make sense for arbitrary bit strings, e.g. copying, moving, input/output operations (e.g. file read and write), applying Boolean functions, shall have formal parameters of this type.

Much of what I have described here already existed in the language IBM PL/I, more than 60 years ago, except that in it only the conversion towards a bit string was explicit, with the built-in function "bit", while the conversions from bit strings to other data types were done implicitly, upon variable assignment.

Like any kind of array, a bit string must have an associated size, so there should be no need to specify it explicitly as a separate parameter.

rfgplk2• 1 month ago

void DoSomething(const void* p, size_t numBytes)

would be something like

template <typename T> void DoSomething (const T& ref) or void DoSomething(const T& ref, size_t numBytes) or C++20-y void DoSomething (const auto& ref)

If the class you're passing in already qualifies a size like member fn, template<typename T> requires requires(T t){ t.size(); } void DoSomething(const T& x){ ... x.size(); }

void DoSomething(const uint8_t* p, size_t numBytes)

This is awful you lose type info irreversibly.

template <typename T> void DoSomething(std::span<T> data)

You can do this but the above examples work just as well.

Or maybe something even more complicated, like this?

template <typename T, std::size_t N> void DoSomething(std::span<T, N> data)

// Or this? template <typename T, std::size_t N> void DoSomething(std::span<const T, N> data)

This is more explicit, not more complicated...

In this way, we still keep the clarity and simplicity of the function invocation: DoSomething(&data, sizeof(data));

Stripping types is not a good idea, especially because you'll run into object lifetime issues _REALLY QUICKLY_. You need to guarantee that the object is trivially copyable.

void-star• 1 month ago

This!

… Is why I picked my name.

jeffbee• 1 month ago

While I sympathize with the aesthetic theme of this post, I warn against the temptation to do what the post proposes, which is to try to compute the checksum of an object represented by void pointer and extent. There are many dangers here, one of which is that the checksum may read uninitialized memory, making the checksum meaningless. Another is that the pointer implicitly converted to void may be a different address depending on the type of the object in the calling function, if the type has multiple base classes. Further, your void reader may be reading derived class data you were unaware of, such that hashing a Base pointer twice yields different results because a member of Derived was placed by the compiler in the tail padding of Base.

In other words, don't do this. C++17 introduced has_unique_object_representations type trait which tells whether it is safe to do this to a given type. It is pretty much always false.

nine_k• 1 month ago

I first wanted to compare the use of void * to the use of a chainsaw. But then I realized that a chainsaw was many more safety features.

ndesaulniers• 1 month ago

I find this point to be generally why C can typically beat C++ in terms of code size; generic functions operating on void* are much less type safe, but the tradeoff is code size. Those template instantiations add up.

gpderetta2• 1 month ago

Making DoSomething a template because span is a template is a non-sequitur.

If DoSomething works with untyped bytes, it should require a std::span<byte> (or const byte if read only). Incidentally the standard provides a convenient as_bytes(std::span<T>)->std::span<byte>; There isn't an as convenient helper to convert a singular object to a span of bytes, but it is easy to write.

As to why one should use span, is that a) it helps making sure that the size travels together with the pointer for some additional safety, b) it is more convenient to work with byte ranges than void ptrs (which do not support pointer math), c) helps a bit communicating intent: in C++ void* are used more often for type erasure than for byte related things.

kstenerud2• 1 month ago

The real question here is: WHY are you passing a blob of memory rather than a struct that uses the type system to describe and enforce what the contents are?

I don't mean dressing up an anonymous pointer, which the author rightly complains about. I mean WHY are you making an API that takes such a pointer to an unknown type to begin with? Whenever you change the structure within that blob, your type checker won't flag that the receiver hasn't been updated to handle it.

Even worse: nothing's stopping you from accidentally passing in the wrong type.

And now you have a SEGV. Or a security hole.

locknitpicker8• 1 month ago

The real question here is: WHY are you passing a blob of memory rather than a struct that uses the type system to describe and enforce what the contents are?

I completely agree. It's particularly egregious when the blogger complains that the complexity and ugliness lies in the type casting to force an incompatible type where it doesn't belong, and use a reinterpret_cast of all things.

This doesn't even feel like a strawman argument anymore. This sounds like a coding horrors entry.

fp643• 1 month ago

I think the article names hashing as a use-case, which I can somewhat still agree. Operations that only depend on the bytes, I guess. But yeah, most things worth saying about this article have been said here already

throwaway274485• 1 month ago

Even then, accepting a uint8_t* would make this intent clearer.

kstenerud2• 1 month ago

Sure, if the function is expected to not treat the data as anything but bytes, then it might be acceptable in narrow circumstances.

But in such a case I'd argue FOR the ceremony, as a way of declaring from the API "The input is a sequence of bytes that I won't treat as anything other than a sequence of bytes", and declaring from each and every call site: "This is not a mistake; we really are 'converting' this struct to a series of bytes for this function to consume".

Then anyone auditing the code knows the intent by the shape of the types, and would quickly flag any typecasting shenanigans within the receiver function.

But even then, hashing a struct will rapidly bring you into the land of dragons and fairies. Abandon all hope if you have floats or UTF-8 (which have multiple representations for the same values).

Far better to remain type-aware if you value your sanity.

xigoi• 1 month ago

Hashing everything based on the byte representation breaks when you have a type where equality does not imply byte equality. Such as… floats (+0 and -0 are equal, but have different byte representation).

scott01• 1 month ago

For type erasure (it’s sometimes useful), custom allocators, I/O, for example.

okanat• 1 month ago

Then pass an uint8_t with size aka a span<uint8_t>.

9rx• 1 month ago

Whenever you change the structure within that blob, your type checker won't flag that the receiver hasn't been updated to handle it.

The relevant type is "blob". There is no further structure. If the function that accepts void* is trying to extract structure out of the blob, there is a bug in that function and the type checker should already catch you trying to extract structure from something that isn't there.

I mean WHY are you making an API that takes such a pointer to an unknown type to begin with?

It's not unknown in any meaningful sense. It is known to be a sequence of 'arbitray' datums of a given length, which is the exact type of input required for the scenario given.

As the article explores, some argue that you should define that sequence with a concrete type, but the article states that it doesn't offer any additional value as is posits that void* already communicates the same. In other words, it suggests that void* is the concrete type for that type already.

DennisL123• 1 month ago

<snark> Arguably the one thing C++ is great at is its type system. Makes total sense to cast it away. </snark>

newsoftheday• 1 month ago

On the other hand, with the “safe and modern” uint8_t prototype, the function call gets more complicated, as you need to add a type cast

To me, that is a feature, not an issue.

girfan• 1 month ago

Partly serious, partly in jest: so type systems are no good?

jurschreuder2• 1 month ago

I would argue that you would not really know what the compiler would be up to with the memory void* points to.

To make sure I would put it in some kind of container.

jurschreuder2• 1 month ago

Nobody uses void* in C++ you will get lectured if you do that.

Ailrk• 1 month ago

Lots of ways to criticize modern c++ but I don’t think void* vs span is a valid one.

Panzerschrek• 1 month ago

  void DoSomething(
   _In_reads_bytes_(numBytes) const void * p,
   _In_ size_t numBytes );

It's an anti-pattern in C++, which causes a lot of bugs and security vulnerabilities. That's why C++ Core Guidelines recommend to use std::span.

Bizwen• 1 month ago

Don’t people remember the real history anymore? The original C did not include the void type, and therefore void* did not exist either. Original C used char* to represent pointers of any type and directly cast char* to double*. You can read the first edition of The C Programming Language from 1978 to learn about this. void and void* were essentially invented based on practices from early C++ and were eventually incorporated into ANSI C.

The beauty and simplicity of the good old C-style void* in C++

Related Stories