This article on Web Components by @leaverou nicely encapsulates a fair few of my feelings about where we are at with ActivityPub / ActivityStreams in the standards process:
> the focus is still on making things _possible_ rather than making them _easy_
Is a fantastic sentence.
It is possible to build a recipe sharing app with ActivityPub. Doing so in a way that interacts nicely with the Fediverse is an unsolved problem.
@thisismissem Yeah I think if ActivityPub was less of a hassle people would code for and thus eventually use more fedi apps. Which is why I think @fedify is a good thing but I am currently turned off by their performance, as happens with JS. But let's be real, Ruby on Rails is no performance powerhouse either...
@dtomvan @fedify you're referring to the recent Ghost article? see here where I debunk the likely cause: https://hachyderm.io/@thisismissem/113240474950357709
Essentially if you autoscale an activitypub server based on open HTTP requests, when significant time is actually spent doing network i/o and the CPU is just idling, then your autoscaling isn't setup right.
@thisismissem @fedify Yeah, I saw that post and fedify's response (with the benchmarks). Those figures look better. Not using a queue is basically like sending someone a message, then doing nothing until they respond, and going on with your day only after they did (if I understand it correctly)... But still I'm not too fond of writing backends in JS. But you are right, these things aren't computationally intensive, so maybe peformance doesn't matter as much.
@thisismissem @fedify Ah, didn't know that. Props for the effort to them. Then I guess they know way better than me which tool in this case is better for the job. Let's see how things turn out and what gets made with it