This is a follow-up to [[P3294R2] (Code Injection with Token Sequences)](https://wg21.link/p3294r2). In that paper, we argued that the right model for code generation in C++ is through token sequences by comparing to other potential models. In short, we believe that code generation in C++ must be C++-shaped, must be able to generate all of C++, and must allow for the full use of C++ in that generation. Hence, token sequences.