{
  "id": 7340220,
  "title": "My coverage print read 13 of 13 on every run. It was incapable of printing anything else.",
  "url": "https://urgent.news/2026/09/14/my-coverage-print-read-13-of-13-on-every-run-it-was-incapable-of",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-09-14T14:58:42.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/mahirhir/my-coverage-print-read-13-of-13-on-every-run-it-was-incapable-of-printing-anything-else-33k9"
  },
  "original_language": "en",
  "account": "One of the crates identifies failure levels within a set, with each entry displaying a code. The system checks how many categories the classifier covers. The worst category showed 13 successes out of 13, which initially seemed reassuring. However, the underlying mechanism was flawed. The numerator and denominator were derived from the same list of arms, causing the fraction to always equal 1, regardless of input or circumstances. This created the illusion of a perfect system, even when an error case arose with no corresponding arm. The solution was to derive the denominator from a separate source - the list of all possible error kinds. This adjustment accounted for cases where an error existed but no matching arm was available. The check now reported the correct coverage percentage, such as 13/14 when a gap appeared in the error kinds. The key takeaway was that coverage numbers only matter when their denominator is sourced independently from their numerator. Relying on the same list for both values creates a false sense of completeness and can mask real issues. An audit system was proposed to systematically check whether the numerator's source differs from the denominator's, flagging any checks that fail this critical test.",
  "summary": "One of the crates classifies a failure into its worst matching category. The category list is an array where each entry carries a code, and a check prints how much of that list the classifier actually covers. OK_WORST_CLASS arms=13/13 Thirteen of thirteen, every run, for as long as the check had existed. I read that as good news for months. It was not news at all. The numerator was built out of…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}