Skip to content

MIGRATION.md

Latest commit

 

History

History
94 lines (78 loc) · 5.17 KB

MIGRATION.md

File metadata and controls

94 lines (78 loc) · 5.17 KB
Raw

Migration / provenance

This repository is a static archive of lambdadays.org, captured 2026-07-08.

Source

lambdadays.org served each edition at https://www.lambdadays.org/lambdadays<year> from a CMS backend. Pages were fully server-rendered (real content in the HTML), so the archive is a faithful static mirror — not a rebuild.

What was captured

  • 12 editions, 2014–2025, each at /lambdadays<year>/index.html.
  • 490 speaker & talk detail pages at /lambdadays<year>/<slug>/ — the bios and talk descriptions linked from each edition (crawled breadth-first to closure). 502 HTML pages in total.
  • 2 original microsites (/static/lambda2014.htm, /static/lambda2015.htm) from the earlier URL scheme, preserved verbatim (images inlined as data URIs).
  • All reachable /static/** assets (CSS, JS, images) and S3-hosted images (lambdadays-prod bucket), downloaded and localized under /static/ and /static/s3/ so the archive survives the source site and its S3 bucket.
  • Inline-style backgrounds — speaker photos, logos and section/venue backgrounds set via style="background-image:url(...)" and <style> blocks (not just <img> tags), all downloaded and localized.
  • Per-edition section theming — on the live site /static/js/theme.js loads an edition-specific theme-<edition>.js that applies most section styling at runtime via jQuery (the pink "Become a sponsor" background, the blue venue block, section colours, speaker-name colours). Relying on that JS is fragile for an archive, so those rules are baked into the static HTML as inline styles + a <style> block, exactly as the JS would set them — the sections render correctly with no JavaScript. The theme files and the images they reference are also captured. (Editions 2014/2023/2024/2025 use theme-2023.)
  • Talk slide decks — presentation PDFs/PPTX linked from speaker pages, including those hosted on S3 (/presentations/, /pdf/).
  • Linked documents (CfP forms, reports, "regulamin" rules) under /static/upload/media/*.pdf.
  • Slide-deck PDFs are recompressed (Ghostscript /ebook, 150 dpi) for size, only where the result is valid and smaller.

What was changed

  • Asset URLs localized — remote /static and S3 image URLs rewritten to local paths; CSS url(...) references followed and downloaded.
  • Cross-edition navigation made relative — links between editions rewritten from absolute www.lambdadays.org/lambdadays<year> to /lambdadays<year>/, so the archive is browsable independent of the domain.
  • Speaker slugs with diacritics recovered — some source landing pages link speaker pages with diacritics dropped (jerzy-mller, 404); the real page uses transliteration (jerzy-muller). Recovered the correct slug from each speaker's display name and rewrote the links (7 fixed on the 2014 edition).
  • Tracking / popup scripts stripped — Google Tag Manager / analytics and GetResponse popup scripts removed.
  • CMS edit-mode link artifacts (/admin_page_elements/edit_block/...) cleaned back to the real destination URL.
  • Genuine external links (YouTube, sponsors, social, Erlang Solutions / Code Sync) left absolute and unchanged.

Known gaps

Everything the source still serves is captured. Verified across all 514 pages (20k+ references): only the following are missing, and each is already dead at the source — nothing recoverable:

  • 2 broken links: jan-machek / jan-macháček (2014 speaker — the detail page 404s at source even at the correct slug; his photo and schedule entry are present, and his bio is in the /static/lambda2014.htm microsite), and one 2022 talk slide PDF (...juliuenigmaelm.pdf, 404 at source).
  • 1 dead speaker photo — the 2018 Nikhil Barthwal headshot (404 at source); it degrades to an empty background, not a broken-image icon.

Sponsor logos (2017 / 2018)

The 2017 and 2018 sponsor/partner logos were 404 at the source S3 bucket. They were restored by following each sponsor's link and sourcing a current logo (from the sponsor's own site, Wikimedia, or the brand's icon), saved under /static/logos/.

  • Restored (18 slots): PTI, Grand Parade (uses the William Hill logo — Grand Parade is part of William Hill), Codegram, ironSource, Microsoft, Sphere Research (now Sphere Engine), IT-Leaders, Geek Girls Carrots, uSwitch, Zimpler, Subvisual, IHS Markit, SlideMight, ElixirFountain, ConfEngine, SPOJ, WITI. Sourced from each sponsor's site, Wikimedia (Microsoft, IHS Markit, ironSource, William Hill), or the Wayback Machine for defunct ones (Geek Girls Carrots, ElixirFountain). Codegram is a low-resolution icon — the only mark still available.
  • Skipped, tile removed: fby.by (2017 & 2018 — now redirects to a Russian-language loans site), and Kraków JS (domain parked/for sale, no logo recoverable anywhere).

The cookie-consent popup that the live site loads was also removed from the archive.

Tooling

Capture/build tooling lives in the hub repo under migration-data/lambda_days/ (build_lambda_days.py, idempotent/resumable: crawl then build). Re-running regenerates this site from the raw page mirror.