The Universal Headers Project

** STATUS: VAPORWARE **

This project distributes a set of C headers (.h files) that are compatible with widely used libcs for various targets.

Instead of copy+pasting N different libc header files, and requiring the use of -I in C compiler invocations, this repository provides cleaned up headers which use the C preprocessor in order to automatically support different targets with the same set of includes.

The main purpose of this project is for simplicity and smaller installation size for the use case of cross-compiling.

In the context of Zig, this project exists to facilitate ziglang/zig#2879. Idea being that the end-result we want to eventually accomplish, is a set of multi-target C headers (this project) plus implementations of the functions in Zig. The C headers from this project would be periodically synchronized into the Zig repository upstream.

The files are in include/. Everything else in this repository exists to aid the creation and maintenance of those files.

Strategy

When files are very different, the idea is to keep the different files in subdirectories and use the preprocessor to #include them as appropriate depending on target information.

When files are very similar, the idea is to edit the files so that different targets share the same files, and the preprocessor is used to contain the differences between targets.

This will require carefully backporting changes if any of the libcs change upstream.

Some preprocessor defines are already present to communicate the target; for example the presence of _WIN32 can be used to detect the Windows operating system. However others are not available; for example there is no preprocessor definition available to detect the difference between gnu or musl ABI. So this project needs to define some custom preprocessor macros that can be used this way.

We may want to look into automating this by parsing .h files and intelligently, using type information, determine what is the same and what is different per target, and then generate the .h files based on this information. Then this repository would have a set of headers as inputs, and a set of headers as outputs. The idea would be that the outputs would be fewer size in bytes because of deduplication, support all the targets without any -I flags needing to be different because of preprocessor usage, and the update process would look like copy+pasting new versions of upstream headers into the input headers, and then re-running the processing step.

This will reduce the effort needed to add support for more libcs, as well as keep the installation size of the headers small, even as the number of supported targets grows large.

Desired Targets

• Mingw-w64 Headers
  • x86_64-windows-gnu
  • i386-windows-gnu
  • aarch64-windows-gnu
  • arm-windows-gnu
• Musl Headers
  • x86_64-linux-musl
  • i386-linux-musl
  • aarch64-linux-musl
  • arm-linux-musl
  • mips64-linux-musl
  • mips-linux-musl
  • powerpc64-linux-musl
  • powerpc-linux-musl
  • riscv64-linux-musl
  • s390x-linux-musl
• glibc Headers
  • x86_64-linux-gnu
  • x86_64-linux-gnux32
  • i386-linux-gnu
  • aarch64-linux-gnu
  • aarch64_be-linux-gnu
  • arm-linux-gnueabi
  • arm-linux-gnueabihf
  • armeb-linux-gnueabi
  • armeb-linux-gnueabihf
  • csky-linux-gnueabi
  • csky-linux-gnueabihf
  • mips-linux-gnu
  • mips64-linux-gnuabi64
  • mips64-linux-gnuabin32
  • mips64el-linux-gnuabi64
  • mips64el-linux-gnuabin32
  • mipsel-linux-gnu
  • powerpc-linux-gnu
  • powerpc64-linux-gnu
  • powerpc64le-linux-gnu
  • riscv64-linux-gnu
  • s390x-linux-gnu
  • sparc-linux-gnu
  • sparcv9-linux-gnu
• MacOS
  • x86_64-macos.10-none
  • x86_64-macos.11-none
  • x86_64-macos.12-none
  • aarch64-macos.11-none
  • aarch64-macos.12-none
• Not associated with any particular implementation:
  • wasm-wasi-musl
  • wasm-freestanding-musl
• FreeBSD
• NetBSD
• DragonFlyBSD
• OpenBSD
• Fuchsia
• Haiku
• Freestanding

License

This project is MIT (Expat) licensed. It will contained derived work from various other licenses:

• Public Domain
• APSL