The whole point of zulip is to be something between a forum with a tree of things that is more async friendly but also be real time enough to get good discussions going for the real time folk.
And I think that forcing that via matrix with it's linear chats would probably hurt things, unless you could have an integration per thread and invited folk based on the stream subscriptions, but that even may be super messy and not as good as just interacting here.
A matrix type bot that didn't pretend to be a full integration but was more a bot that just pushed interesting new things to you like notifications of new threads, mentions, etc.
If you then say, "there is this stream of X and anyone in the federation can create new channels/topics" then that's another new layer of conflict resolution across the federated set of homeservers.
Otherwise, what happens if homeserver X and homeserver Y both create a "Reflex FRP" topic in the Haskell stream? Who wins? How do the other homeservers decide?
Yeah I guess a room per topic as a bridge would work, but new topics would be pretty painful to keep track of. I dunno if I'd use it because I'd miss the "all messages" stream that's so good for keeping tabs on the whole thing.
Extending the Quote functionality to allow for threaded messaging. This could be a simple 'Quote 2.0'-style like WhatsApp where the quoted text is hyperlinked and when clicked makes you scr...
Extending the Quote functionality to allow for threaded messaging. This could be a simple 'Quote 2.0'-style like WhatsApp where the quoted text is hyperlinked and when clicked makes you scr...
Has anyone setup a matrix bridge for this zulip? I think it would be a nice thing to have!
I love matrix, but I think that would ruin the benefits of Zulip!
The whole point of zulip is to be something between a forum with a tree of things that is more async friendly but also be real time enough to get good discussions going for the real time folk.
And I think that forcing that via matrix with it's linear chats would probably hurt things, unless you could have an integration per thread and invited folk based on the stream subscriptions, but that even may be super messy and not as good as just interacting here.
Yeah, it would be amazing if matrix had the topic feature. Otherwise I see the problem.
A matrix type bot that didn't pretend to be a full integration but was more a bot that just pushed interesting new things to you like notifications of new threads, mentions, etc.
That could be fun, but I dunno.
Well that bot exists.
But someone would need to host it.
Yeah, I wish that matrix wasn't at the whim of async unfriendly linear chat rooms too! But having topics may make their federated stuff way harder.
You, think? What is the problem with topics and federation?
It adds new primitives the the distributed set of actions that are being federated. And synapse is already slow enough.
Maybe the rewrite will fix things, but I'd be hesitant to add things to the protocol until they can prove that this one scales.
Fair enough
why not a channel per topic?
Because channels aren't currently federated. There are local and under the control of a single homeserver.
If you then say, "there is this stream of X and anyone in the federation can create new channels/topics" then that's another new layer of conflict resolution across the federated set of homeservers.
Otherwise, what happens if homeserver X and homeserver Y both create a "Reflex FRP" topic in the Haskell stream? Who wins? How do the other homeservers decide?
since zulip isn't federated I would assume that the bridge would only allow sending messages to existing topics
Oh, we're confusing the bridge vs topics being part of matrix here.
Yeah I guess a room per topic as a bridge would work, but new topics would be pretty painful to keep track of. I dunno if I'd use it because I'd miss the "all messages" stream that's so good for keeping tabs on the whole thing.
I'm pretty sure it could be implemented on tom of the existing matrix protocol, if it had good support on the client side
I would love that
Apparently that is/was the plan https://github.com/vector-im/riot-web/issues/2349#issuecomment-249391252
cool
Still planned ! https://github.com/vector-im/riot-web/issues/2349#issuecomment-625745647