skip to content
Posts
-
Finding redundant Rust tests with per-test coverage
How per-test LLVM coverage, set comparisons, and assertion review find redundant Rust tests, and why matching coverage alone cannot justify deletion. -
Rust got left behind on Supabase, so I built supabase-rp
TypeScript gets generated types and typed queries on Supabase. Rust's usual answer, SQLx, goes around the API and RLS. supabase-rp closes the gap. -
Laya beat my Rust game. So did random picks.
Laya, a small 421M-parameter AI model, won 0 of 10 floors in my Rust game. Once Rust did the moving, it won 10 of 10. So did a bot picking at random. -
Code Review Is Becoming Team Alignment
AI makes code cheaper to write, so review has to move from inspecting every line to keeping the team aligned on intent, risk, and rollout. -
Writing Software You Don't Run
Patterns for building CLIs, protocols, and customer-hosted agents when old data, old binaries, and old users keep existing. -
Patterns From Five Years in Blockchain
The closing post on five years of blockchain work: boundaries, bytes, compatibility, tooling, and the cost of making systems agree. -
Irys: Storage, Consensus, and Execution
A technical map of consensus, EVM execution, storage chains, and the protocol engineering behind Irys Programmable Data. -
Axelar-Solana: A Cross-Chain Protocol on a 1232-Byte Budget
Porting Axelar general message passing to Solana meant fitting 27 secp256k1 signatures and multi-kilobyte payloads into a 1232-byte transaction ceiling. -
The Builder Era
Five years in blockchain, part 1. Provably-fair NFTs, renting without forced returns, and an AMM that tried to manage liquidity for passive LPs. -
Why Are You Not Claudemaxxing?
From mass-dismissing AI coding tools to managing 5 Claude sessions on a walk. A prompt school origin story -- and your first useful AI win.