on a whim i had a possibly funny idea. changing the post-increment operator in c from ”++” to ”+++” and renaming the language c+++. how hard could it be? (spoiler: not very).

from my ongoing reading of the wonderful Crafting Interpreters, a token in programming languages is simply a small sequence of characters that mean something in the language. for example: (, !=, ., ++, and return are various kinds of tokens in C. these tokens get built by the tokenizer, or lexer, or scanner… (there are slight differences).

since ++ is a standalone token, theoretically all i would need to do is have the tokenizer recognize ”+++” as whatever structure the ++ token is in clang. and yeah, that was literally it. it didn’t take too long for me to stumble upon the tokenizer (clang/lib/Lex/Lexer.cpp) and scroll around until i saw where most single character characters were being tokenized.

  case '+':
    Char = getCharAndSize(CurPtr, SizeTmp);
    if (Char == '+') {
      CurPtr = ConsumeChar(CurPtr, SizeTmp, Result);
      Kind = tok::plusplus;
    } else if (Char == '=') {
      CurPtr = ConsumeChar(CurPtr, SizeTmp, Result);
      Kind = tok::plusequal;
    } else {
      Kind = tok::plus;
    }
    break;

Lexer.cpp (line 4113)

with a massive leap of faith, i ended with something that seemed like it could work.

  case '+':
    Char = getCharAndSize(CurPtr, SizeTmp);
    if (Char == '+') {
      Char = getCharAndSize(CurPtr, SizeTmp);
      CurPtr = ConsumeChar(CurPtr, SizeTmp, Result);
      if (Char == '+') {
        CurPtr = ConsumeChar(CurPtr, SizeTmp, Result);
        Kind = tok::plusplus;
      }
    } else if (Char == '=') {
      CurPtr = ConsumeChar(CurPtr, SizeTmp, Result);
      Kind = tok::plusequal;
    } else {
      Kind = tok::plus;
    }
    break;

Lexer.cpp (modified)

4 hours of compiling and 50 GB of build data later, it worked!

i’d probably like to play around with clang again later cause of how easy this change was.