Gregory's Server
Is there an #ActivityPub platform built with #Java ?
(No, I'm not building one, and neither should you)
|
3 posts total
Yesterday I implemented a #JsonLd algorithm myself (instead of delegating to Titanium)! No, it doesn't mean I'll implement all the other ones myself --- but, I'm very tempted to do so. The algorithm (although I've skipped two short non-important steps for now) is only 15~20 lines! For comparison, the same algorithm (which is the document loafer one btw) is 150~250 lines in Titanium. #Kotlin is awesome!  @laxla Do you really need it? JSON-LD is not required for federation, and may even be harmful because some fediverse servers don't produce valid JSON-LD (at least that's what I heard - I don't do JSON-LD processing myself) |
@laxla
Why shouldn't I? I would love to have a #JVM (ideally #kotlin) implementation. The performance would be much better than ruby, and JVM-based data options such as Cassandra, Elasticsearch and kafka would all outperform the Mastodon choices. You could run it all on a single node and scale it fairly big, I suspect. I gave it a shot back in 2017 but I couldn't wrap my head around the spec. Now that there are better tools to learn the spec, I don't have the time :/
The one I've seen mentioned most often is Smithereens.
@laxla https://github.com/grishka/Smithereen