Would be helpful for most new users to understand how Lemmy works and how the different hosts interact with each other in a basic way.
Would be helpful for most new users to understand how Lemmy works and how the different hosts interact with each other in a basic way.
A detail that is useful to keep in mind, as it can explain some edge cases, but which isn’t critical to actual day-to-day usage, is that whenever you’re viewing a community hosted on another server, you’re not viewing the content on that server. Instead, you’re viewing a mirror of it hosted on your local server.
This means there may be comments or posts viewable on the community’s host server that you’re not seeing yet, because they haven’t been passed along to the instance you’re using yet. The remote community needs to actively forward new posts and replies to instances that are following it.
This is also why if you follow a remote community from a new instance – that is, you trigger the mirroring – you won’t necessarily get the whole backlog of posts from the community. Just like how if you subscribe to a magazine, you only get future print editions (they don’t send you their entire back catalogue), when an instance subscribes to a remote community, it only receives future content.
Interesting, do you think this approach will limit how big Lemmy can get? For example, if a server wanted to subscribe to ten communities that are the size of large subreddits, how much data would that be? How much would it cost to maintain a server that could handle that?
Generally speaking, text is cheap to transmit and store. It’s images and video that could be a real issue.
But ultimately, content on small instances may end up being somewhat ephemeral. Developers and admins may want to look into ways to earmark significant posts so that they don’t end up in the dustbin, but 90%+ of what gets posted to social media isn’t actually worth saving long term anyway.
Only text is mirrored, images and video are hosted on the instance where they were posted, so overall it’s really cheap to store all of that, and even more so if the load is distributed across many instances.