A mistake corrected, told in full
The real story of a failure in automatic sends, and what it taught about building well.
Full text in preparation. The main ideas:
- Told as a story, not as a list of changes.
- How to investigate in order of real probability: first prove the right version is running, then the logic, then the environment.
- Why every send has to say how many went out, how many failed and why.
