Hacker News got a post that was utterly delightful today, showing a REPL for Z80 assembly. It's not a new project - it's eight years old - but it doesn't matter, because holy cow: a Z80 REPL.
A Z80 REPL! You enter a Z80 instruction, it gets assembled and slotted into memory at the program counter, it executes and you see what happened to registers and memory!
You get completion - like a REPL! - and protection against trivial loops, and other features. Given that the Z80 was one of the processors that powered a rather significant portion of the microcomputer revolution, this is brilliant, nostalgic, and would have blown minds back in the day when kids were still being warned not to mess with their CRTs.
Consider what doing this used to be: you'd write assembly code, look up the opcodes, load your assembler if you had one (or manually convert the code if not, which was much more "interesting")... then you'd execute your code, only to find you'd made a terrible mistake, and reload.
And today we can just LD A, 10h and see what actually happens, immediately.
The biggest thing here is that this is developer experience. The feedback loop used to be labyrinthine - have an idea, encode it, enter, execute, discover, re-encode, re-enter, re-execute, re-discover...
But now, with a REPL, you can literally have the idea, try it out, and discover. The activation energy for doing things has dramatically lowered.
It's the same thing we see with AI: the loops are shortened, because it's so easy to try things, but the difference between this and using an AI is that this is illustrating the actual low-level bits to the programmer instead of hiding them behind an LLM. You get to learn more here!
The author, Kenta Cho, has been writing small games for 40 years, driven by a love of making games he wants to play himself, embracing constraints, tiny programs, custom tooling, and rapid experimentation, and he's published an ebook online: The Joys of Small Game Development.
The Z80 REPL, along with his other work, is absolutely fascinating, and it's well worth considering: even if you never intend to write a byte of Z-80 code, the idea is worth stealing: make experimentation cheap enough and people experiment.