By this I sort of mean like, in the case of Lemmy for example, something like a built-in Lemmyverse of sorts.
It wouldn’t be pulling in data yet from other instances, but it would enable seeing that their communities were out there so you could then tell your server you’d like to subscribe/follow/etc. their community/channel/account.
Tbh despite having been around the fediverse for awhile, I’m not sure how else ActivityPub servers were supposed to have these different things found to start with.
This feels like an impossible combination, where your local instance doesn’t have data from other instances, but also knows that they are out there. When a user searches their instance for something, the answer comes from local storage, in the same way that google responds by looking at its local cache of a webpage, not the webpage itself.
That’s kind of what I was thinking may be the case, but I’m not sure if I’m asking this well enough or if I may be misunderstanding ActivityPub.
It’s not clear to me how, without communication/searching outside of an ActivityPub instance, it would ever find other ActivityPub instances to connect to and communicate with.
This feels like an impossible combination, where your local instance doesn’t have data from other instances, but also knows that they are out there
I think they are suggesting a sort of tiered system of federating communities, where local instances store data on the existence of all communities, but not necessarily their entire contents.
Yeah, I know what you mean. I originally typed out a reply based on that assumption, but then re-read the post, and decided against it, in case I was assuming the wrong thing.
Currently, if you bring in a remote community, your instance fetches the details of: the community; its moderators (including their avatars and banners); the instance the community is on; the admins of that instance (including their avatars and banners); and the last 50 posts.
If you separated that into tiers, tier 1 could just be the community details (inc. the sidebar), and the instance name, and you could maybe skip the rest until someone subscribed to it. I don’t think you’d want to bring in every available community, even for tier 1, because of the 29k communities listed on lemmyverse.net, at least 20k are completely dead (about 6k of that is just spam from a disgruntled lemmy.world user). The advantage of this is that a local user could find the community based on a local search. They’d be greeted by an empty-looking community though, so they’d have to know that subscribing to it would be worthwhile (probably by visiting it on its host instance, although apps can make that process difficult by being a bit too clever for their own good).
There’s an Issue for PieFed that been raised about using something like lemmyverse to ‘know’ about remote communities. I’m ‘freamon’ in that discussion, but neither me or PieFed’s developer have pursued it much further.
Not really. You can utilise a Mastodon client that allows you to subscribe to remote instances. Another option is to utilise a relay and subscribe to an instance
It wouldn’t be pulling in data yet from other instances, but it would enable seeing that their communities were out there
It would have to pull some data, but it could be much less than federating every community.
I think it would be great if, as soon as two servers become aware of each others existence, they automatically federate each others lists of communities, and possibly the sidebar info.
Federating the rest of the data would happen only if someone subscribes to a community from another instance, same as it is now.
