About

The rule

No change is published without a quote that appears literally in the project's own release notes. The quote is checked by substring match before anything is stored: whitespace is normalised, nothing else. A paraphrase is rejected even when it is accurate, because a paraphrase that passed would defeat the point of having a quote at all.

That catches invention. It does not catch omission — a breaking change that was in the notes and never got picked up. Nothing automatic catches that, so changes are scored against a hand-annotated set of upgrades where the right answer is known, and nothing is published until every blocker-severity change in that set is found.

The reason for being this careful is asymmetry. A reader warned about something that is not there loses an hour. A reader not warned about something that is loses a weekend, and then rightly never trusts the site again.

What a page means

• Releases in a span — checkable facts, straight from the project.
• "Not extracted yet" — nobody has looked at this span. Not a clean bill of health.
• A listed change — quoted from the source, with a link to it.

Where the data comes from

Release history from each project's own releases, support cycles fromendoflife.date. Nothing is scraped that has an API, requests are conditional so an unchanged source costs one round trip, and the collector identifies itself.

Release notes are never republished. They belong to the projects that wrote them. What appears here is a summary written by this project, a short quote for verification, and a link to the original. If you maintain a project and would rather not appear,open an issue and it comes out.

What it will not become

It will not tell you an upgrade is safe. It tells you what the project said changed, and where it said it. The decision stays yours, because the person who has to fix it at 2am is you.

Data generated 2026-11-02 from 2fc3266· source and catalogue