Go to file
Andrew Kelley f1eafe4ebb fix bug in RNG example 2016-01-06 03:53:30 -07:00
c_headers parseh: add c header files 2015-12-08 17:51:59 -07:00
cmake cmake needs to be taught a lesson in debian paths 2015-12-06 23:19:03 -07:00
doc support unknown size arrays 2016-01-06 01:28:58 -07:00
example fix bug in RNG example 2016-01-06 03:53:30 -07:00
src codegen: fix lshr/ashr not looking at int sign 2016-01-06 03:52:49 -07:00
std support unknown size arrays 2016-01-06 01:28:58 -07:00
test add unknown size array test 2016-01-06 01:30:11 -07:00
.gitignore link against LLVM 2015-08-05 15:23:15 -07:00
CMakeLists.txt provide std.zig and add it to import paths 2015-12-10 17:42:47 -07:00
LICENSE add license 2015-08-05 16:22:18 -07:00
README.md cleanup readme 2016-01-04 04:44:32 -07:00

README.md

zig lang

An experiment in writing a low-level programming language with the intent to replace C. Zig intends to be a small language, yet powerful enough to write optimal, readable, safe, and concise code to solve any computing problem.

Porting a C project to Zig should be a pleasant experience - every C feature needs a corresponding Zig feature which solves the problem equivalently or better.

Zig is not afraid to roll the major version number of the language if it improves simplicity, fixes poor design decisions, or adds a new feature which compromises backward compatibility.

Goals

  • Completely compatible with C libraries with no wrapper necessary.
  • In addition to creating executables, creating a C library is a primary use case. You can export an auto-generated .h file.
  • Do not depend on libc unless explicitly imported.
  • Provide standard library which competes with the C standard library and is always compiled against statically in source form.
  • Generics so that one can write efficient data structures that work for any data type.
  • Ability to run arbitrary code at compile time and generate code.
  • No null pointer. Convenient syntax for dealing with a maybe type so that null pointer is not missed.
  • A type which represents an error and has some convenience syntax with regards to resources.
  • Defer statement.
  • Eliminate the need for configure, make, cmake, etc.
  • Eliminate the need for header files (when using zig internally).
  • Tagged union enum type.
  • Resilient to parsing errors to make IDE integration work well.
  • Ability to mark functions as test and automatically run them in test mode. This mode should automatically provide test coverage.
  • Friendly toward package maintainers.
  • Ability to declare dependencies as Git URLS with commit locking (can provide a tag or sha1).
  • Include documentation generator.
  • Shebang line OK so language can be used for "scripting" as well.

Current Status

  • Have a look in the examples/ folder to see some code examples.
  • Some language features available such as loops, inline assembly, expressions, literals, functions, importing other files.
  • Only Linux x86_64 is supported.
  • Only building for the native target is supported.
  • Optimized machine code that Zig produces is indistinguishable from optimized machine code produced from equivalent C program.
  • Zig can generate dynamic libraries, executables, object files, and C header files.
  • The binaries produced by Zig have complete debugging information so you can, for example, use GDB to debug your software.

Building

Debug / Development Build

If you have gcc or clang installed, you can find out what ZIG_LIBC_DIR should be set to (example below).

mkdir build
cd build
cmake .. -DCMAKE_INSTALL_PREFIX=$(pwd) -DZIG_LIBC_DIR=$(dirname $(cc -print-file-name=crt1.o))
make
make install
./run_tests

Release / Install Build

Once installed, ZIG_LIBC_DIR can be overridden by the --libc-path parameter to the zig binary.

mkdir build
cd build
cmake .. -DCMAKE_BUILD_TYPE=Release -DZIG_LIBC_DIR=path/to/libc/dir
make
sudo make install