Replay.web is cool

I’ve been trying to preserve as much of Placemark now that it’s open-source. This has been a mixed experience: some products were really easy to move away from, like Northwest and Earth Class Mail. Webflow was harder to quit. But replay.web came to the rescue, thanks to Henry Wilkinson at web.recorder.

Now placemark.io is archived, but nearly complete and at feature-parity, but costs next to nothing to maintain. The magic is the wacz format, which is a specific flavor of ZIP file that is readable with range requests. From the geospatial world, I’ve been thinking about range requests for a long time: they’re the special sauce in Protomaps and Cloud Optimized GeoTIFFs. They let you use big files, stored cheaply on object storage like Amazon S3 or Cloudflare R2, but lets browsers read those files incrementally, saving the browser time & memory and saving you transfer bandwidth & money.

So, the placemark.io web archive is on R2, the website is now on Cloudflare Pages, and the archive is little more than this one custom element:

<replay-web-page source="https://archive.placemark.io/placemark%202024-01-19.wacz" url="https://www.placemark.io/"></replay-web-page>|

This is embedding replayweb.page. Cool stuff!