8 notes
A conceptual essay on where Mochi MEP-42 should sit on the curve from "compile slowly, run fast" (LLVM -O3) through "compile and run at medium speed" (Cranelift, B3) to "compile instantly, run okay" (copy-and-patch, Sparkplug). Includes a back-of-envelope budget derived from MEP-17 and MEP-23 bench numbers.
The published material a Mochi engineer should keep open while implementing MEP-42 phase 1. Cooper/Torczon 3rd ed for the canonical theory, Nystrom for a hands-on bytecode compiler walkthrough, Appel for the verified-compiler-curious, plus 2024-2026 course materials covering naive backends.
The most actively researched mid-tier compiler backend of the 2020s. ISLE for instruction selection, proof-carrying code for Wasm-sandbox memory accesses, and VeriISLE for verified instruction-lowering rules. Where the "design and implementation of a portable codegen framework" line of work currently lives.
The Triton dialect (OpenAI, GPU codegen), the IREE dialect (Google, ML compiler), Mojo's MLIR-based KGEN compiler, and the broader trend of language frontends as MLIR dialects. The "if we wanted to be ambitious" backend story for Mochi.
Recent foundational work from POPL and its co-located CPP (Certified Programs and Proofs) workshop on verified compilation, secure calling conventions, and packet-filter codegen. Less directly applicable to a "naive emitter" than PLDI, but instructive for the verification story MEP-42 may eventually want.
Recent work from the flagship PL conference on lightweight backends, baseline JIT design, superoptimization for stack bytecode, and library composition that sidesteps heavyweight compilation.
PLDI 2024-2025 codegen, POPL 2024-2025 compiler papers, MLIR dialects 2026, Cranelift design, compiler textbooks 2026, compile-time vs runtime tradeoff.
How a mathematical paper or article is organized so that readers can follow the main ideas.