Welcome back to my lair! Recently I’ve revived a project that some may know: C².

C² is a superset of C, designed as basically a nicer-to-use C version.

Some of the features/changes it has compared to C:

These are planned.

• Built-in error types (Zig-like) with error unions and void!-like (void can be any type) function return types.
• Namespaces (not like C++).
• Better enums based in enum class from C++, as C’s enums declare global constants that can collide in name, meaning most C programs must use some prefix for the enum, which is much less nice than just using MyEnum.Whatever instead MyEnum_Whatever as a constant.
• Completely structural typing where field names don’t matter and types must only have fields/items of the same type to be considered compatible.
• Better include system that ditches headers for a module-based approach.
• Better memory management that ditches raw mallocs and free for allocator structures, built-in keyword-based freeing (free keyword) and defer for automatic end-of-block cleanup.
• More types quadruple-precision floats¹, built-in from day one booleans (C got booleans pretty late as _Bool in C99, and there are still programs that use <stdbool.h> for booleans or int directly).
• Safer types by being more strict with casts, forcing uncasted values to be the exact same type (meaning that you can’t assign an uint to an ulong witho ut casting) and clearer type-specific literals (100.0f for floats, 100u for unsigned integers of any size, etc.).
• VM-based for faster iteration, packages supported and you can create binaries out of the runtime + the bytecode.

What is implemented right now #

For now, I’ve made a lexer that runs pretty fast (100k iterations in ~44μs, or ~0.000044 seconds), an error printing system:

and a test suite:

Right now I am working in the parser, which will be probably finished soon and I will start with the real deal: compilation!.

Links #

• GitHub repo