https://mattermost.com/marketplace/mattermost-matrix-connector/
This is what RocketChat were doing since about 2022. Although their documentation suggests they are working on full Matrix support inside the RC server itself, due sometime this month;
Post
https://mattermost.com/marketplace/mattermost-matrix-connector/
This is what RocketChat were doing since about 2022. Although their documentation suggests they are working on full Matrix support inside the RC server itself, due sometime this month;
@strypey the linked source code was last touched 2 years ago. Is it really new or even actually working? Seems a bit strange 🤔
@HarHarLinks
> the linked source code was last touched 2 years ago
It's a simple bit of sticky tape between a MM and Synapse server. Maybe doesn't need active development?
Also lack of updates fits my hypothesis that it's a stopgap, while they work on a full Matrix implementation in MM server itself.
> Is it really new
Only to me it seems : P
> even actually working?
Haven't tested, no idea. Any MM hosters in the house?
(1/?)
@controlfreak
> What's better than just one resource abyss?
The performance cost of Synapse has always been overstated. People would join dozens of high-traffic public rooms, and wonder why their homeserver was doing so much work. There were real technical chokepoints, but many of these have been progressively solved;
https://matrix.org/blog/2018/12/20/synapse-0-34-0-released/
https://matrix.org/blog/2020/11/03/how-we-fixed-synapse-s-scalability/
Matrix 2.0 brings further improvements, like Sliding Sync;
@strypey bad past experience aside, never keen to try again given their warm embracing stance to take money to gargle "law enforcement" and gubment spying.
(1/?)
@controlfreak
> their warm embracing stance to take money to gargle "law enforcement" and gubment spying
I doubt we're not going to agree on whether this is justifiable or not. But I'm happy to argue the toss if you're up for it.
(2/?)
Having said all that, running Synapse alongside an existing chat silo server is always going to use more resources than using the protocol directly in the chat server. I'm guessing that as with RocketChat, the supported bridge is a stopgap to provide the UX of Matrix interoperation. While they work on implementing the protocol internally. Which is why I made the comparison : )