On Sat, Jan 5, 2019 at 6:35 PM, j.r <j...@wiuwiu.de> wrote:
Why do you think this? I think that wouldn't that much complicated...
Because replication in distributed systems is an absurdly complex topic by itself and this complexity will inevitably lead to implementation bugs which is even harder to debug than bugs in multi-threaded system. It's somewhat possible to design this from scratch in relatively straightforward way, but with all that XMPP baggage I'm very skeptical. Going a bit offtopic: we still have a lot of problems with message delivery in XMPP, for example, we have no protection against server failures during message delivery or we don't have reliable delivery in presence of s2s failures. Yet, you want to add another source of message delivery failures. _______________________________________________ Standards mailing list Info: https://mail.jabber.org/mailman/listinfo/standards Unsubscribe: standards-unsubscr...@xmpp.org _______________________________________________