And I was curious about integrating with more than RSS/Atom because a lot of valuable things are not available that way. So far, that includes GitHub (issues, PRs, releases, and notifications), Pinboard bookmarks, Email (creates addresses on the fly for you to use), ConvertKit, and an API. I had Twitter integrations but had to retract them because of recent policy, cost, and operational issues with their API.
I linked directly to where you can try it out for free with no credit card or password creation.
Each page of feed browsing takes 50ms or less on the server side, making it snappy, to stay in the flow state while browsing and triaging. One of the design tenets was to be "fast for all feeds at all timestamps" — digging into the past needs to be just as fast as recent items.
I built it on DynamoDB and Redis, with a Django frontend that calls backend Rust services which do most of the heavy lifting. The asynchronous system is backed by SQS and a custom workflow engine that triggers Linux binaries (Docker and Lambda are also used where I found that more appropriate).
I'm looking forward to blogging about how it works behind-the-scenes, and some lessons learned.
The main site link: https://www.yupdates.com
A blog post with more background: https://www.peakscale.com/yupdates/
Thanks for looking!
No comments yet.
