• B Language grammar - here
• Language reference - here

---

Blog series: Credence

The compiler works in 3 stages:

• The Lexer, Parser first-pass, built with an LALR(1) grammar and parser generator in python that interfaces with C++ via pybind11

• An IR I've called Instruction Tuple Abstraction or ITA - a linear 4-tuple set of platform-agnostic instructions that represent program flow, scope, and type checking

• The target platforms - currently x86_64, arm64, and z80

---

There are a few differences between the compiler and B specification, namely:

• Roughly strict typing with type inference
  • Vectors (arrays) may be non-homogeneous
  • The trivial vector is strongly typed
  • Uninitialized variables are set to an internal null type
• Compile-time out-of-range boundary checks on vectors and pointer arithmetic
• No undefined behavior
• Boolean "truthy" coercion for all data types in conditionals
• GOTO and labels are not supported, use control structures
• Support for C++ style comments
• Logical and bitwise operators behave more like C
• Uses C operator precedence
• Switch statement condition must always be enclosed with ( and )
• Binary operators may not be used directly after the = operator
• Constant literals must be exactly 1 byte

Usage

Credence :: B Language Compiler
Usage:
  Credence [OPTION...] positional parameters

  -a, --ast-loader arg   AST Loader [json, python] (default: python)
  -t, --target arg       Target [ir, syntax, ast, arm64, x86_64, z80]
                         (default: ir)
  -d, --debug            Dump symbol table
  -o, --output arg       Output file (default: stdout)
  -h, --help             Print usage
      --source-code arg  B Source file

./credence --help

Targets

The default compile target is currently ITA:

Example:

B Code:

main() {
  auto *a;
  auto c, i, j;
  extrn unit;
  c = unit;
  a = &c;
  i = 1;
  j = add(c, sub(c, i)) - 2;
  if (c > i) {
    while(j > i) {
      j--;
    }
  }
  c = 0;
}

str(i) {
  extrn mess;
  return(mess[i]);
}

add(x,y) {
  return(x + y);
}

sub(x,y) {
  return(x - y);
}

unit 10;

mess [3] "too bad", "tough luck", "that's the breaks";

ITA:

__main():
 BeginFunc ;
    LOCL *a;
    LOCL c;
    LOCL i;
    LOCL j;
    GLOBL unit;
    c = unit;
    _t2 = & c;
    a = _t2;
    i = (1:int:4);
    _p1 = c;
    _p3 = c;
    _p4 = i;
    PUSH _p4;
    PUSH _p3;
    CALL sub;
    POP 16;
    _t3 = RET;
    _p2 = _t3;
    PUSH _p2;
    PUSH _p1;
    CALL add;
    POP 16;
    _t4 = RET;
    _t5 = _t4;
    j = (2:int:4) - _t5;
_L6:
    _t9 = c > i;
    IF _t9 GOTO _L8;
_L7:
    c = (0:int:4);
_L1:
    LEAVE;
_L8:
_L10:
_L12:
    _t13 = j > i;
    IF _t13 GOTO _L11;
    GOTO _L7;
_L11:
    j = --j;
    GOTO _L10;
 EndFunc ;


__str(i):
 BeginFunc ;
    GLOBL mess;
    RET mess[i] ;
_L1:
    LEAVE;
 EndFunc ;


__add(x,y):
 BeginFunc ;
    _t2 = x + y;
    RET _t2;
_L1:
    LEAVE;
 EndFunc ;


__sub(x,y):
 BeginFunc ;
    _t2 = x - y;
    RET _t2;
_L1:
    LEAVE;
 EndFunc ;

---

Test suite

make test

---

Installation

Note: $PYTHONHOME must be set to an installation that has chakram installed.

Ubuntu

sudo apt-get update
sudo apt-get install -y gcc-10 llvm-20 valgrind clang-20 iwyu python3-dev cppcheck clang-tidy pipx
# Inside the repository:
echo 'eval "$(register-python-argcomplete pipx)"' >> ~/.profile
source ~/.profile
cmake -Bbuild -DCMAKE_BUILD_TYPE=Debug -DUSE_SANITIZER="Address;Undefined" -DCMAKE_EXPORT_COMPILE_COMMANDS=ON
cmake --build build

MacOS

brew update
brew install coreutils include-what-you-use llvm@20 cmake python3 pyenv
# Inside the repository:
cmake .. -DCMAKE_BUILD_TYPE=Debug -DUSE_SANITIZER="Address;Undefined" -DCMAKE_EXPORT_COMPILE_COMMANDS=ON
cmake --build build

Windows (mingw/msys)

pacman -S git wget mingw-w64-x86_64-clang mingw-w64-x86_64-gcc mingw-w64-x86_64-ninja mingw-w64-x86_64-cmake make mingw-w64-x86_64-python3 autoconf libtool
# Inside the repository:
# Note: iwyu and the -fsanitizers do not work in mingw
cmake -Bbuild -DCMAKE_BUILD_TYPE=Debug -DCMAKE_EXPORT_COMPILE_COMMANDS=ON
ninja

Installing chakram

git submodule update --init --recursive
cd python/chakram
pipx install poetry # or similar
poetry install
# Be sure to use pyenv or similar
make install

---

Dependencies

Note: These are installed automatically via CPM and cmake.

• simplejson++ - Lightweight memory safe json library
• chakram - LALR(1) parser generator and Lexer
• cxxopts - Lightweight commandline parser
• matchit - Pattern matching
• cpptrace - Stack traces for runtime errors until C++23
• eternal - constexpr lookup tables
• pybind11

License

Apache 2 License.