Skip to content

post-conference.md

Latest commit

 

History

History
63 lines (46 loc) · 3.87 KB

post-conference.md

File metadata and controls

63 lines (46 loc) · 3.87 KB
Raw

After the conference

The hand-off window between the closing keynote and the moment the edition becomes part of the archive. Three operational threads run in parallel; they can be tackled in any order.

Thread 1 — Recordings

YouTube uploads typically lag the event by 4–8 weeks. Once the post-production team finishes a batch, link the recordings to talk pages.

Step Doc
Sync recordings from a YouTube playlist docs/recordings.md
Override / patch individual matches docs/recordings.md#recordings-yaml
Re-run social-card generation if new keynote artwork dropped make sponsor-pages is sponsor-only; for talk social cards run uv run python utils/social_card_img_gen.py

A talk with a linked recording gets the YouTube embed on the talk page, a recording_available: yes flag, a VideoObject JSON-LD record, and an entry in the video sitemap.

Thread 2 — Transcripts

Auto-generated from the recording audio, then imported into the talk's contents.lr. Once present, transcripts unlock per-talk full-text search and inflate the searchable corpus considerably (~14k lines for a typical edition).

Step Doc
Import a batch from a local directory docs/transcripts.md
Format spec (Speaker 1 / Speaker 2 diarization, panel speaker_map) docs/transcripts.md#format

The importer is idempotent — re-running it on already-imported talks refreshes the field rather than appending.

Thread 3 — Front page after the event

The site has one always-conference front page (no recap/inactive mode, no activate/disable switch). When the conference closes, do not flip the homepage to a recap state — instead:

  • Publish the gratitude recap ("year in numbers", photos, thank-you) as blog posts and link them from the homepage programme_status band (databags/frontpage/sections/programme_status.yaml) for a short while.
  • Move the closing edition under /archive/{year}/ (see the cutover steps below) so its content lives in the archive.
  • Once the next edition's dates lock, update the front-page configs (databags/frontpage/sections/*.yaml, databags/frontpage/config.yaml) and databags/site_state.yaml to the next conference. The persistent archive funnel-CTA reads site_state.yaml and will point archive traffic at the new conference automatically.

Front-page config details: docs/landing_pages.md.

Cutover to the archive

When the next edition's site goes live, the current edition moves under /archive/{year}/. Sequence:

  1. Final Pretalx pull. Run the importer one more time to capture late changes to bios or session metadata. See docs/pretalx.md.
  2. Recordings + transcripts complete. Everything you want crawlable should be in by this point — once a page lives under /archive/, it inherits EventCompleted and eventStatus flips in JSON-LD.
  3. Move content. Rename content/talks/ to content/archive/{year}/talks/ and content/speakers/ to content/archive/{year}/speakers/. Create content/archive/{year}/contents.lr from a previous edition as a template.
  4. Re-generate derived content. 
    uv run python utils/migrate_pretalx_slugs.py    # slug URLs + Pretalx-code redirects
uv run python utils/generate_tracks.py          # per-track index pages
make build                                       # everything else (sitemaps, search, redirects)
    Both generator scripts are idempotent — they're safe to re-run on already-migrated content.
  5. Verify. Pick three random talks, check the slug URL renders, the Pretalx code URL redirects, and the speaker page lists them. Crawl the new edition with the local static server (see docs/search.md) to confirm internal links resolve.

The archive layout and conventions are documented in docs/archive.md.