llvm-mirror | Fork of llvm with experimental commits and workarounds | Compiler library

As the comments above say, you're comparing the libstdc++ version of all_of with the libc++ version which you copied into your code, and the libstdc++ version is faster.

As for why it's faster: std::all_of calls std::find_if_not, which calls __find_if_not. This function negates the predicate and calls __find_if, passing it the iterator category as an additional parameter.

__find_if has two overloads: One for input iterators, which is similar to the implementation you showed above. The other is for random access iterators. This overload can calculate the length of the range, which allows it to unroll the loop and perform four predicate calls in each iteration. This is the reason it can work faster on random-access containers such as std::vector.

The pcm-files contains "half baked" information about a module. The file extension does not have to be pcm and i think Visual Studio for example uses .ifc as extension. (And a difference with Visual Studio is that it outputs the ifc-file and compiles the object file (.obj) at the same time, while clang does this in two separate steps)

The pcm file is storing the information about the module file in a format so that the compiler can easily import it or compile it to other object files.

The command from your link surprises me, I think that -S in the example is compiling to assembly, so i am surprised that it is listed in that clang repo that you linked to.

For every function you use, consult the C++ reference to find out which header you should include.

The reference for fabs says that you have to include .

The problem seems to be that uninitialized global variables are somehow optimized out and not added to the symbol table.

A quick work around to ensure that the variable gets added to the symbol table is to initialize it with an "Undefined value".

The following code allows to do such an initialization with the c++ api

The compilation itself succeeds even if the file has syntax errors. Consider the following example:

The iterator-constructor for std::span you mention was proposed in P1394R4. According to libc++ C++2a Status, this paper has not yet been implemented in libc++.

libc++ implements P0122R7, where there is that pointer-constructor (replaced by the iterator-constructor in P1394R4).

P1872R0 is also related: span should have size_type, not index_type.

An implementation is allowed to add noexcept to a non-virtual function if it wont ever throw an exception, see [res.on.exception.handling]/5 of the C++17 standard (draft N4659).

However an implementation is not allowed to add constexpr to a function. See [constexpr.functions]/1. See also LWG issue 2013.

std::condition_variable::condition_variable() is specified as neither constexpr, nor noexcept, but there are no circumstances under which it must throw an exception. See [thread.condition.condvar].

So, noexcept is fine, but constexpr is not. However, functions being marked constexpr that shouldn't be, is a common non-conformance. For example GCC declares math functions deliberately constexpr although they are not supposed to be.

I made a stupid and common mistake, "always redirect stdout along with stderr"

like this