Link-rot triage, and three kinds of redirect
Every outbound link checked. Nothing is dead - but reading the results properly mattered more than the headline, because most of the redirects were not what they looked like.
Masterdoc §7 asked how much of what this archive points at is at genuine risk. Everything here is twenty-odd years old and privately hosted, so the honest answer needed measuring rather than assuming. A script now collects every http link out of the data files, checks each one, and reports.
The headline is good: of 49 distinct links, none are dead and none are unreachable. Three return a refusal rather than a failure - MobyGames and the Fandom wiki bot-check, and GitHub rate-limited the twenty-sixth request in a row - which is a different finding from a broken page and is counted separately.
Eleven redirects, three kinds
- Six that are not redirects at all
- Every GOG storefront link resolved to a /fr/ URL, because the machine running the check is in France. Rewriting those would have hard-coded one visitor's locale into the data for everyone. Left alone.
- Four that are a missing slash
- Three HeavenGames roots and a Discord invite redirect to a canonical form. Harmless, but a hop is a hop, so the trailing slashes are now in the data.
- One that is a real move
- The Caesar 3 Heaven forum has moved from /cgi-bin/forums/ to /cgi-bin/caeforumscgi/. The old URL still redirects today; the day it stops, three links on this site break. Updated, and verified at the new path.
The script reports rather than deletes, and that is deliberate. A 403 usually means a bot check, not a dead page, and quietly dropping a good link because a CDN dislikes scripts would lose more than it saves. Only a genuine 4xx or 5xx fails the run.