llvm-energy | Static energy consumption for LLVM IR | Compiler library

References can't bind to objects with different type directly. Given const int& s = u;, u is implicitly converted to int firstly, which is a temporary, a brand-new object and then s binds to the temporary int. (Lvalue-references to const (and rvalue-references) could bind to temporaries.) The lifetime of the temporary is prolonged to the lifetime of s, i.e. it'll be destroyed when get out of main.

Clang is wrong.

[dcl.fct.def.general]

2 The type of a parameter or the return type for a function definition shall not be a (possibly cv-qualified) class type that is incomplete or abstract within the function body unless the function is deleted ([dcl.fct.def.delete]).

That's pretty clear I think. A deleted definition allows for an incomplete class type. It's not like the function can actually be called in a well-formed program, or the body is actually using the incomplete type in some way. The function is a placeholder to signify an invalid result to overload resolution.

Granted, the parameter types are more interesting in the case of actual overload resolution (and the return type can be anything), but there is no reason to restrict the return type into being complete here either.

You can find an authoritative answer in microsoft/TypeScript#16 (wow, an old one); it is specifically explained in this comment:

That's because of document.all [...], a quirk that gets special treatment in the language for backwards compatibility.

The TLDR is that loads which miss all levels of the TLB (and so require a page walk) and which are separated by address unknown stores can't execute in parallel, i.e., the loads are serialized and the memory level parallelism (MLP) factor is capped at 1. Effectively, the stores fence the loads, much as lfence would.

The slow version of your insert function results in this scenario, while the other two don't (the store address is known). For large region sizes the memory access pattern dominates, and the performance is almost directly related to the MLP: the fast versions can overlap load misses and get an MLP of about 3, resulting in a 3x speedup (and the narrower reproduction case we discuss below can show more than a 10x difference on Skylake).

The underlying reason seems to be that the Skylake processor tries to maintain page-table coherence, which is not required by the specification but can work around bugs in software.

For those who are interested, we'll dig into the details of what's going on.

I could reproduce the problem immediately on my Skylake i7-6700HQ machine, and by stripping out extraneous parts we can reduce the original hash insert benchmark to this simple loop, which exhibits the same issue:

According to the C++17 standard (11.3.6 Default arguments)

9 A default argument is evaluated each time the function is called with no argument for the corresponding parameter. A parameter shall not appear as a potentially-evaluated expression in a default argument. Parameters of a function declared before a default argument are in scope and can hide namespace and class member name

It provides the following example:

Edited: For people who use Cocoapods, this answer might be useful: https://stackoverflow.com/a/69384358/587609

I also faced this issue, and it seems that there is a known issue on Xcode 13 as mentioned in this document: https://developer.apple.com/documentation/Xcode-Release-Notes/xcode-13-release-notes

Swift libraries depending on Combine may fail to build for targets including armv7 and i386 architectures. (82183186, 82189214)

Workaround: Use an updated version of the library that isn’t impacted (if available) or remove armv7 and i386 support (for example, increase the deployment target of the library to iOS 11 or higher).

If your app is for iOS 11 or higher, one of the libraries should be modified to target iOS 11 or higher (e.g., my app is for iOS 12 or higher).

For example, I am using GRDB.swift, and its minimum iOS version is 10.0. There was a discussion as an issue of this repo, and I followed that comment to solve this issue as follows:

1. Fork the repository
2. Change Package.swift to modify the minimum iOS version like:

The code shown is valid (all C++ Standard versions, I believe). The similar restrictions are all listed in [reserved.names]. Since read is not declared in the C++ standard library, nor in the C standard library, nor in older versions of the standard libraries, and is not otherwise listed there, it's fair game as a name in the global namespace.

So is it an implementation defect that it won't link with -static? (Not a "compiler bug" - the compiler piece of the toolchain is fine, and there's nothing forbidding a warning on valid code.) It does at least work with default settings (though because of how the GNU linker doesn't mind duplicated symbols in an unused object of a dynamic library), and one could argue that's all that's needed for Standard compliance.

We also have at [intro.compliance]/8

A conforming implementation may have extensions (including additional library functions), provided they do not alter the behavior of any well-formed program. Implementations are required to diagnose programs that use such extensions that are ill-formed according to this International Standard. Having done so, however, they can compile and execute such programs.

We can consider POSIX functions such an extension. This is intentionally vague on when or how such extensions are enabled. The g++ driver of the GCC toolset links a number of libraries by default, and we can consider that as adding not only the availability of non-standard #include headers but also adding additional translation units to the program. In theory, different arguments to the g++ driver might make it work without the underlying link step using libc.so. But good luck - one could argue it's a problem that there's no simple way to link only names from the C++ and C standard libraries without including other unreserved names.

(Does not altering a well-formed program even mean that an implementation extension can't use non-reserved names for the additional libraries? I hope not, but I could see a strict reading implying that.)

So I haven't claimed a definitive answer to the question, but the practical situation is unlikely to change, and a Standard Defect Report would in my opinion be more nit-picking than a useful clarification.

It is allowed to initialize a char array with a string if the array is at least large enough to hold all of the characters in the string besides the null terminator.

This is detailed in section 6.7.9p14 of the C standard:

An array of character type may be initialized by a character string literal or UTF−8 string literal, optionally enclosed in braces. Successive bytes of the string literal (including the terminating null character if there is room or if the array is of unknown size) initialize the elements of the array.

However, this also means that you can't treat the array as a string since it's not null terminated. So as written, since you're not performing any string operations on b, your code is fine.

What you can't do is initialize with a string that's too long, i.e.:

In C, a void* can be implicitly converted to any T*. As such, making NULL a void* is entirely appropriate.

But that's profoundly dangerous. So C++ did away with such conversions, requiring you to do most pointer casts manually. But that would create source-incompatibility with C; a valid C program that used NULL the way C wanted would fail to compile in C++. It would also require a bunch of redundancy: T *pt = (T*)(NULL);, which would be irritating and pointless.

So C++ redefined the NULL macro to be the integer literal 0. In C, the literal 0 is also implicitly convertible to any pointer type and generates a null pointer value, behavior which C++ kept.

Now of course, using the literal 0 (or more accurately, an integer constant expression whose value is 0) for a null pointer constant was... not the best idea. Particularly in a language that allows overloading. So C++11 punted on using NULL entirely over a keyword that specifically means "null pointer constant" and nothing else.