Skip to content

Clean up legacy blog post link rot (~30 dead links across 2015-2018 posts) #36

@jmxpearson

Description

@jmxpearson

Summary

The lychee link checker added in PR #32 surfaced ~30 dead links across older _posts/ (2015-2018). The site-health.yml workflow currently excludes the entire blog directory (--exclude-path _site/blog) so these don't gate PRs.

Sample of what's broken

  • http://people.duke.edu/~sni/* (5+ deleted personal-page assets)
  • http://nasher.duke.edu/faculty/... (site reorganized)
  • https://dibs.duke.edu/scholars/elizabeth-johnson and dibs.duke.edu/events/... (pages moved/deleted)
  • https://bassconnections.duke.edu/project-teams/... (old project page)
  • https://medschool.duke.edu/research/.../basic-science-day (moved)
  • https://www.neuro.duke.edu/people/faculty/eva-naumann (moved)
  • https://livestream.com/naturalsciences/Cafe (livestream URLs gone)
  • http://www.upenn.edu/pennnews/news/michael-platt-... (Penn news article moved)
  • http://research.vtc.vt.edu/people/read-montague/team (moved)
  • http://www.sdsc.edu/~bourne/ (gone)
  • http://emuseum.nasher.duke.edu/... (returns 400 Bad Request; API changed)
  • 3-4 [ERROR] connection failures (changlab.net, kellyraechi.com, etc. — possibly defunct)

Full list available by running lychee --config ./lychee.toml --no-progress --root-dir "$(pwd)/_site" _site locally without the blog exclude (or just see the lychee report from PR #32 commit 59c3abb).

Acceptance

For each error, decide one of:

  • Update: replace with the page's current URL
  • Archive: swap to an Internet Archive snapshot (https://web.archive.org/web/<timestamp>/<url>)
  • Remove link: keep the surrounding prose, drop the <a> / [text](url) markup
  • Remove sentence: rare; only if the linked content was the entire point

Then drop --exclude-path _site/blog from .github/workflows/site-health.yml so blog posts get ongoing checking.

Notes

  • This is editorial work, not infrastructure. Could be split into one issue per post if helpful.
  • Consider doing this after the dibs-web01 migration so we know the link checker is reliably catching the relevant cases.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions