A reader tells us a page may be wrong. Sometimes we can settle it the same day against a manufacturer document. Often we cannot settle it at all without the equipment in front of us. This register is what happens to the second kind.
An open erratum does not change the page. It publishes the fact that a question exists, dated, attached to the exact step, and visible to the next reader — while the specification on the page stays exactly as its source has it. This repository never publishes a value it cannot trace to a manufacturer or standards source, and a reader report is not such a source. Publishing the disagreement instead is the same rule it already follows when two sources disagree with each other.
Four states, after the way the IETF handles corrections to a published RFC:
- reported — a credible report, not yet verified. It renders on the page it concerns, above the step it concerns.
- verified — a tier-1 source confirmed it. The page is corrected and the change is listed in the corrections log.
- held for update — credible, and not being acted on now. Recorded so that it is not lost, and it still renders on the page.
- rejected — checked, and either wrong or already covered. It stays here with its reason.
Nothing is ever removed from this register, and a reader's words are quoted as they were written. An erratum is closed by adding a resolution to it, never by deleting the report. A page with an open erratum on it is not retired.
No erratum has been recorded yet.