February 15, 2026
Real-time or real hype?
Show HN: Copy-and-patch compiler for hard real-time Python
Devs call it “insane” — a fast, on‑time Python that could rival C tools
TLDR: Copapy promises fast, predictable “on‑time” Python for hardware by compiling to machine code, but it’s still a proof of concept with no hardware I/O yet. The community is split between wild excitement about replacing existing speed‑boost tools and cautious calls for write‑ups, demos, and proof this really works
Imagine Python that shows up on time, every time. That’s the pitch behind Copapy, a new “copy‑and‑patch” trick that turns Python‑style code into lean machine code for robots, drones, and other hardware. The crowd’s reaction? Loud. One fan yelled “THAT’S INSANE,” while another said it could rival the usual speed‑ups people use with Python — think NumPy, Numba, and even Cython (those are popular tools to make Python faster or closer to C). The vibe: this could be the “PyTorch for embedded” — easy to write, fast to run, and predictable in timing.
But here’s the twist: it’s still a proof of concept. There’s no hardware input/output yet, which sparked a mini‑reality check. Some asked for a deep write‑up and real demos before the hype train leaves the station, while others cracked jokes about “Python paying rent on time” and finally getting “C‑level speed without the tears.” Fans loved that it targets common chips (x86 and ARM), avoids big dependencies, and even does automatic differentiation (think: math that helps software fine‑tune itself) — all while using a tiny C “runner” to handle the boring stuff. Bottom line: hype levels are high, skeptics want receipts, and everyone’s refreshing for that first hardware demo.
Key Points
- •Copapy is a Python framework for deterministic, low-latency real-time computation with automatic differentiation.
- •It uses a copy-and-patch compiler to produce optimized, statically typed, memory-safe machine code with minimal runtime errors.
- •Current architecture support includes x86_64, ARMv6, ARMv7, and AArch64; porting requires only a C compiler.
- •Generated binaries avoid system calls and external libraries (e.g., libc), easing deployment on RTOS or bare metal; a C runner manages execution and unified relocations.
- •The project is a proof of concept without hardware I/O yet; ongoing work includes SIMD array stencils, Thumb support for ARM-M MCUs, and constant regrouping.