Seven times my test harness reported success while measuring nothing
Last week a stranger reproduced my benchmarks. Over four days he found six defects. Every single one of them had been reporting success. That is the part worth writing down. Not that I had bugs — everyone has bugs. That my instrumentation, six separate times, returned a number that looked fine and meant nothing. A crash is loud. A harness that prints 86.1% when it measured the wrong thing is not,…
A stranger discovered six bugs in my benchmarks over a four-day period. Six separate times, my instrumentation reported success numbers that were meaningless. One bug involved a path length limit on Windows, causing the benchmark to treat many tables as non-existent. Another bug resulted from a comment that had gone unnoticed, causing certain picks to be displaced.
An enum missing a value led to silent failures when branching on a specific field. A check with power zero by construction returned 0 of 52 questions, despite comparing against the same value. An aggregate was never printed, and the held-out evaluation failed to include all necessary schemas. Finally, a fix that converted a crash into silence by attaching an in-memory database did not change the measurement.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
