May 28, 2026
Baked goods, bugs, and big opinions
Durable Execution the Hard Way
One coder says you can build a crash-proof job runner by hand, and the crowd is equal parts impressed and exhausted
TLDR: A developer published a learn-by-building guide for making software that can recover after crashes, using only Go and Postgres. Readers are split between calling it a refreshingly honest masterclass and joking that putting everything in the database is beautiful chaos waiting to happen.
A new guide called Durable Execution the Hard Way has dropped, and the pitch is wonderfully nerdy: build your own “never lose progress” job system from scratch using Go and Postgres, step by step, with almost no extras. In plain English, it’s a tutorial for making long-running programs remember where they were before a crash. The author frames it as a back-to-basics learning exercise, ending with a bare-bones but working workflow engine — basically a traffic controller for big, multi-step computer tasks.
But the real fireworks are in the reactions. One camp is shouting “finally, show us the guts!” and praising the guide for not hiding behind glossy tools and buzzwords. These readers love that it’s opinionated, minimal, and honest about being more about understanding than convenience. The other camp is already clutching its pearls at the phrase “entirely in Postgres,” treating it like a dare, a confession, or both. The hottest debate is whether this is a brilliant teaching tool or the opening scene of someone’s future production disaster.
And yes, the jokes wrote themselves. Commenters are basically doing the classic “the database is now the app” routine, while others are admiring the chaotic charm of offering bug reporters a baked good as a reward. The biggest laugh? The author’s deadpan note that AI didn’t write the prose — all mistakes are handmade. That line alone had people acting like they’d found the indie-artisan version of software tutorials.
Key Points
- •The article is a guide to building a durable execution engine from scratch using Go and Postgres.
- •It defines durable execution as incremental checkpointing of function state so execution can resume after failure.
- •The guide is aimed at developers who want to understand workflow engines like Hatchet and Temporal or build their own.
- •Its lessons progress from task-queue basics to durable event logs, non-determinism tracking, and durable tasks.
- •The guide implements durable execution entirely in Postgres and distinguishes retries, replays, and forking as separate behaviors.