The key idea here seems to be to use function pointer's type to enforce type safety rather than using the data "handle" type (that is often found in implementations inspired by Sean Barrett's strechy_buffers).
> One annoying thing about C is that it does not consider these two variables to have the same type
C23 solves that too: https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3037.pdf
Supported by latest GCC and Clang, but not by MSVC.
Author here. Not quite. The key idea is about using a union to associate type information with a generic data type. Type casting a function is not the only way to use that type information. I discuss that as well as the C23 changes in the footnotes and the "typeof on old compilers" section.