User avatar
jer @nyquildotorg@gts.nyquil.org
1y
I hope the situation where setting up a fediverse instance on a domain "burns" that domain forever gets rectified at some point. I would really like to just use my nyquil.org hostname, but still be able to change platforms at some point. For now I used gts.nyquil.org, assuming that I'm probably going to eventually want to play with bonfire or iceshrimp.NET or some other platform, and can have them all running on their own subdomains until a time when I want to migrate to one of them.
2
1
0
0
User avatar
🪨 @Varpie@peculiar.florist
1y
@nyquildotorg What do you mean by "burning" a domain?
1
0
0
0
User avatar
jer @nyquildotorg@gts.nyquil.org
1y
@Varpie
Once an ActivityPub server has been federated using a domain, it can't ever be used again for another fediverse instance because any instance that has ever connected to it will refuse to talk to it until that instance's owner purges its record of the previous instance using that domain. (And that basically never happens.)
1
0
0
0
User avatar
🪨 @Varpie@peculiar.florist
1y
@nyquildotorg Really? Why? It's all ActivityPub, why would the other servers care about which software is serving it?
1
0
0
0
User avatar
jer @nyquildotorg@gts.nyquil.org
1y
@Varpie
It's about the encryption keys and URL paths that differ between implementations. They all do things slightly differently.

Part of it is trying to make sure someone doesn't snap up an expired domain and suddenly have access to all the previous owner's followers.
1
0
0
0
User avatar
🪨 @Varpie@peculiar.florist
1y
@nyquildotorg Encryption keys should be reusable (the owner should have access to private keys), and there are standards regarding endpoints discoverability (basically everyone uses Webfinger for instance), so that should be solvable with cache invalidation. I think it's more a matter of having a proper migration process than some fundamental incompatibility.
⭐1
1
0
0
1
User avatar
jer @nyquildotorg@gts.nyquil.org
1y
@Varpie
Yeah, I agree. It is "just" a cache invalidation issue caused by a lack of defined process, that is solvable, should every platform decide it's important enough to solve it.

That's what I'm complaining about lol.
1
0
0
0
User avatar
jer @nyquildotorg@gts.nyquil.org
1y
@Varpie
I also wish migration was a better process, but that doesn't help this issue much, since most instances that go away just die when the domain expires. There isn't a defined mechanism for self-destruct or timeout for an instance that has had no activity for a certain period of time.
1
0
0
0
User avatar
🪨 @Varpie@peculiar.florist
1y
@nyquildotorg True, right now when an instance dies the admins of federated instances have to manually remove it, and there is no instructions on whether they should keep the content from that instance or delete/tumbstone it... Definitely something to look into, it could be worth create a FEP
⭐1
0
0
0
1