toycpp: A toy C++ compiler

I'm interested in implementing a compiler for some subset of the C++ language, as a learning exercise, inspired by Tsoding's recreational programming.

The compiler's goal is to produce X86_64 assembly.

Inspirations/resources

• Tsoding's recreational programming series
• Bisqwit's "Creating a compiler" series
• The FASM 1.73 manual
• en.cppreference.com, which I access through DevDocs
• The combined Intel IA-32/64-bit Software Developer's Manual

For information on parsing:

• GLR Parser on Wikipedia
• The GNU Bison manual, particularly 5. The Bison Parser Algorithm
• "(Right-Nulled) Generalised LR Parsing" by Jeff Smits
• "Generalised LR parsing algorithms" by Giorgios Robert Economopoulos (website, pdf)

Development and Usage

1. Install fasm, git, a C++ compiler supporting C++17 and GNU Make:

   sudo apt install git fasm gcc-c++ make

2. Clone the project and compile toycpp using make:

   git clone https://github.com/PracticallyNothing/toycpp
   cd toycpp
   make -j4

3. Run toycpp on a C++ file - this will produce a file called executable in the current directory.

   ./toycpp test/add.cpp

4. Run executable - voila!

   ./executable

Unsupported stuff

Parens around function parameter names

This is to avoid the most vexing parse - C (and C++ by extension) technically allows you to put parentheses around the names of function parameters:

// These two are equivalent.
int main(int  argc,  const char**  argv);
int main(int (argc), const char** (argv));

I have no idea what the use of this is and I've never seen it used anywhere. Skipping support for this makes parsing other things easier.