spiffytech / Brian Cottingham

I also went a step further and automatically synced my bsky follows <-> RSS reader subscriptions, rather than having to manually add RSS feeds for people I want to follow.

Overall I'm quite happy with how it went. The bsky integration was relatively easy, and felt like it was built to do exactly what I was trying to do.

View full thread

Except for how there's no official way to turn a post ID into a link to the bsky website, and the maintainers think that's kind of a philosophically invalid ask in the first place, so you have to chop up the at:// URI string. Feels hacky.

I was able to skip caching because bsky gives me everything I need in a single feed query per author, so I can't get any more efficient while serving fresh data. bsky has a firehose I could have used to detect new content, but for now I just refetch when I get a request from my RSS reader.

View full thread

This makes common things things a little clunky to use, but that's the only big negative I had with the API. Otherwise it was a smooth experience.

One you know what kind of record you're holding, the data structures are straightforward and intuitive. The data needs minimal postprocessing.

Strangely, the lib includes TypeScript type definitions, but values aren't type-annotated. You have to already know a value's type and cast it. Then you get types for all of its immediate properties. Except for child properties whose types can vary, which you (again) have to know & cast.