My motivation for moving out invite-notify is primarily that I want the freedom to not load it, which doesn't seem ideal from a maintenance perspective, but I have tried to test it.
The reason for the rather ugly two-stage solution is that invites can be rejected at the target end, and notifying half the network about an invite that doesn't actually happen seems wrong. This approach has the disadvantage that, if the inviter or invitee quits at the wrong moment, the notification can get lost. I think that's less bad.
My motivation for moving out invite-notify is primarily that I want the freedom to not load it, which doesn't seem ideal from a maintenance perspective, but I have tried to test it.
The reason for the rather ugly two-stage solution is that invites can be rejected at the target end, and notifying half the network about an invite that doesn't actually happen seems wrong. This approach has the disadvantage that, if the inviter or invitee quits at the wrong moment, the notification can get lost. I think that's less bad.