mnutt has quit [Remote host closed the connection]
mnutt has joined #sandstorm
mnutt has quit [Remote host closed the connection]
mnutt has joined #sandstorm
mnutt has quit [Remote host closed the connection]
mnutt has joined #sandstorm
xet7 has quit [Ping timeout: 260 seconds]
amenonsen has quit [Read error: Software caused connection abort]
amenonsen has joined #sandstorm
mnutt has quit [Remote host closed the connection]
mnutt has joined #sandstorm
blowfist has quit [Ping timeout: 248 seconds]
blowfist has joined #sandstorm
blowfist has quit [Ping timeout: 252 seconds]
blowfist has joined #sandstorm
yarmo has quit [Ping timeout: 248 seconds]
yarmo has joined #sandstorm
yarmo2 has joined #sandstorm
yarmo has quit [Ping timeout: 260 seconds]
yarmo2 is now known as yarmo
mnutt has quit [Remote host closed the connection]
mnutt has joined #sandstorm
mnutt has quit [Ping timeout: 252 seconds]
cwebber has quit [Read error: Connection reset by peer]
mnutt has joined #sandstorm
<jfred>
ocdtrekkie: Agreed on wanting something on sandstorm for the fediverse. That'd take some pretty deep integration though, right?
<ocdtrekkie>
Yes and no. To do it right, yes. To do it badly, not at all.
<ocdtrekkie>
I think you could technically give it IpNetwork, have it KeepAlive, and technically, you might be able to run it? But your fedi name might also be 300 characters long?
<jonesv>
ocdtrekkie: is it the same kind of issues as for Matrix?
<jfred>
I was imagining usernames might look like @jfred@<long id>.sandstorm.terracrypt.net
<jfred>
Conveniently I think Mastodon will try and find users your instance knows about in search given only the local part
<jfred>
Less conveniently it will sometimes show the full name in the UI, and it might count against the character limit for people mentioning you...
mnutt has quit [Remote host closed the connection]
mnutt has joined #sandstorm
mnutt has quit [Remote host closed the connection]
mnutt has joined #sandstorm
mnutt has quit [Remote host closed the connection]
mnutt has joined #sandstorm
<ocdtrekkie>
I'm not sure we have any clear documentation on where all the hangups are for either ActivityPub or Matrix. abliss knows a lot of it though probably.
<abliss[m]>
I knew the hangups for matrix and could probably remember them... i don't think i ever tried activitypub
<ocdtrekkie>
It might be a little easier? I don't know. The DNS/server identity stuff will probably be similarly frustrating.
cwebber has joined #sandstorm
<ocdtrekkie>
Also, with current events being as they are, the lack of visibility into grain CPU and memory use might be pretty problematic. The fediverse has been murdering servers lately.
<TimMc>
It would probably be fine for single-user instances.
<jonesv>
Are Mastodon servers as inefficient (if I may say it like this 😅) as the Matrix homeservers?
<jonesv>
(I would be curious to compare an IRC server to a Matrix one, someday. And then also to e.g. Mastodon)
mnutt has quit [Remote host closed the connection]
mnutt has joined #sandstorm
mnutt has quit [Ping timeout: 246 seconds]
mnutt has joined #sandstorm
<ocdtrekkie>
The big problem is that Mastodon caches a LOT of stuff, everything it sees, basically.
mnutt has quit [Ping timeout: 260 seconds]
mnutt has joined #sandstorm
<jfred>
I think you'd probably want something a bit more lightweight than Mastodon anyway, if you're following the Sandstorm pattern of a grain being a small unit within whatever system you're using (probably an AP account per grain would make the most sense IMO)
<jfred>
Epicyon, maybe
<ocdtrekkie>
That would be Sandstorm-y but present some major downsides for ActivityPub
<ocdtrekkie>
Each grain would probably be seen like a different server, wouldn't it? And each grain would be caching all that wonderful federated data.
<ocdtrekkie>
I agree Mastodon is too hefty to make sense though.
<jfred>
It would, but I bet that's often the case anyway (my Sandstorm instance for example is almost entirely single-user) - and for a Sandstorm user, it may be more attractive to create your own grain rather than use someone else's even if they do support multiple users since it's so easy
<jfred>
One big problem that I can see though is that AP inboxes are meant to be long-lived, so e.g. grain backups and restores would not work at all if they resulted in a different randomly-generated endpoint each time one is created
mnutt has quit [Remote host closed the connection]
<jfred>
I think you'd need a portable profile type thing like what Spritely has planned for that to make sense
<ocdtrekkie>
I feel like we eventually need a good system for services at well-known endpoints for narrow cases like federation and stuff.
<ocdtrekkie>
Also, like, mail input is handled by Sandstorm and then handed out to grains. I wonder if ActivityPub should be something similar.
mnutt has joined #sandstorm
<jfred>
Oh, here's a thought: what if, when creating a grain that relied on well-known endpoints, you permanently assigned an ID to that grain in a cryptographically verifiable way - like you include a signed document somewhere in the grain metadata that gets included in a backup. Then if the user restores a backup of that grain, the new grain gets the same ID
<ocdtrekkie>
It'd be the API endpoint that'd need to be consistent, I assume.
<ocdtrekkie>
Restoring webkeys
mnutt has quit [Ping timeout: 260 seconds]
<jfred>
Right - I'm thinking that ID would be part of the domain for things like federation (which webkeys seem to do now)
catern has quit [Remote host closed the connection]