Clang

Background

Starting in 2005, Apple made extensive use of LLVM in a number of commercial systems, including the iPhone software development kit (SDK) and integrated development environment (IDE) Xcode 3.1.

One of the first uses of LLVM was an OpenGL code compiler for OS X that converts OpenGL calls into more fundamental calls for graphics processing units (GPU) that do not support certain features. This allowed Apple to support the entire OpenGL application programming interface (API) on computers using Intel Graphics Media Accelerator (GMA) chipsets, increasing performance on those machines. For GPUs that support it, the code is compiled to exploit fully the underlying hardware, but on GMA machines, LLVM compiles the same OpenGL code into subroutines to ensure continued proper function.

LLVM was intended originally to use GCC's front end, but GCC turned out to cause some problems for developers of LLVM and at Apple. The GCC source code is a large and somewhat cumbersome system for developers to work with; as one long-time GCC developer put it, "Trying to make the hippo dance is not really a lot of fun".

Apple software makes heavy use of Objective-C, but the Objective-C front-end in GCC is a low priority for GCC developers. Also, GCC does not fit smoothly into Apple's IDE. Finally, GCC is licensed under GNU General Public License (GPL) version 3, which requires developers who distribute extensions for, or modified versions of, GCC to make their source code available, whereas LLVM has a BSD-like license which does not force users to release their source code changes when publishing compiled binaries of those changes.

Apple chose to develop a new compiler front end from scratch, supporting C, Objective-C and C++. This "clang" project was open-sourced in July 2007.

Design

Clang is intended to work atop LLVM. The combination of Clang and LLVM provides most of the toolchain, to allow replacing the full GCC stack. Because it is built with a library-based design, like the rest of LLVM, Clang is easy to embed into other applications. This is one reason why most OpenCL implementations are built with Clang and LLVM.

One of Clang's main goals is to provide a library-based architecture, to allow the compiler to be more tightly tied to tools that interact with source code, such as an integrated development environment (IDE) graphical user interface (GUI). In contrast, GCC is designed to work in a classic compile-link-debug cycle, and integrating it with other tools is not always easy. For instance, GCC uses a step called fold that is key to the overall compile process, which has the side effect of translating the code tree into a form that looks unlike the original source code. If an error is found during or after the fold step, it can be difficult to translate that back into one location in the original source. Also, vendors using the GCC stack within IDEs use separate tools to index the code, to provide features like syntax highlighting and autocomplete.

Clang is designed to retain more information during the compiling process than GCC, and to preserve the overall form of the original code. The goal of this is to make it easier to map errors back into the original source. The error reports offered by Clang are also aimed to be more detailed and specific, as well as machine-readable, so IDEs can index the output of the compiler during compiling. Modular design of the compiler can offer source code indexing, syntax checking, and other features normally associated with rapid application development systems. The parse tree is also more suitable for supporting automated code refactoring, as it directly represents the original source code.

Clang is a compiler for only C-like languages, including C, C++, Objective-C, Objective-C++, OpenCL, and CUDA. For other languages, including Java, Fortran, and Ada, LLVM remains dependent on GCC or another compiler frontend. In many cases, Clang can be used or swapped out for GCC as needed, with no other effects on the toolchain as a whole. It supports most of the commonly used GCC options.

Performance and GCC compatibility

Clang is designed to be highly compatible with GCC. Clang's command-line interface is similar to and shares many flags and options with GCC. Clang implements many GNU language extensions and enables them by default. Clang implements many GCC compiler intrinsics purely for compatibility. For example, even though Clang implements atomic intrinsics which correspond exactly with C11 atomics, it also implements GCC's __sync_* intrinsics for compatibility with GCC and libstdc++. Clang also maintains ABI compatibility with GCC-generated object code. In practice Clang can often be used as a drop-in replacement for GCC.

Clang's developers aim to reduce memory footprint and increase compilation speed compared to competing compilers, such as GCC. In October 2007, they report that Clang compiled the Carbon libraries well over twice as fast as GCC, while using about one-sixth GCC's memory and disk space. However, as of 2011 this was not a typical result. As of mid-2014, Clang won more than a third of the benchmarks, with GCC winning most.

While there are still a few tests where performance of a Clang-compiled program lags behind performance of the GCC-compiled program, by large factors (up to 5.5x), it has been reported that Clang "continues to be under very active development," and a hope was expressed for further improvement.

Status history

This table presents only significant steps and releases in Clang history.