"Decoding the MIPS R5900 instructions in each function Translating those instructions to equivalent C++ code Generating a runtime that can execute the recompiled code The translated code is very literal, with each MIPS instruction mapping to a C++ operation." It sounds like a MIPS interpreter that gets statically unrolled.

Yes, it's like the result of unrolling a MIPS interpreter, but there never was an actual MIPS interpreter.

I thought the point of the Futamura projection was that there was actually partial evaluation happening, i.e. you take a real interpreter and specialize it in some automated fashion. That's what makes it interesting.

But I could well be wrong about the naming. It doesn't really matter what it's called if we're all clear about what's actually happening.