Since I recently set up a decently powerful homelab, with a bunch of services running on docker behind traefik, I thought it would have been cool to try out selfhosting a matrix homeserver for personal use (at the beginning it will be just me, potentially it will extend to some friends and/or a small organization of around 20 people working together). At first I was thinking about going with Dendrite, but then I’ve seen it still doesn’t support Threads, nor Matrix 1.5 API and that despite the announcement of a couple of months ago its repo hasn’t yet be moved to the new element-hq GitHub as it happened with Synapse. This made it seem kinda like a “2nd class citizen” compared to Synapse, and therefore made me think if the latter would have been a better option to selfhost. And then I bumped into Conduit.rs so now I have 3 option to be undecided about!

Therefore, here I am asking for your preferences and advices: which is your favourite one / which one would you recommend for my use-case?

  • ninjan@lemmy.mildgrim.com
    link
    fedilink
    English
    arrow-up
    7
    ·
    1 year ago

    +1 on the question. I’m also more for Dendrite since it solves the resource hogging of Synapse. For me the limiter is that the bridges are built for Synapse and will “likely” work on dendrite. If I didn’t have three small kids I’d be up for helping test but I don’t have that kind of time right now.

    • thepaperpilot@incremental.social
      link
      fedilink
      arrow-up
      2
      ·
      1 year ago

      I’d love to move over as well, but for me the deal breaker is SSO. I want it to make my server (with various services) to have its own identity provider which controls access to the whole server

    • MentalEdge@sopuli.xyz
      link
      fedilink
      arrow-up
      2
      ·
      1 year ago

      Unless a server/bridge implements the spec incorrectly, any bridge should work on any server, as they all talk using the appservice API.

      That said, Synapse is the safe option, and for personal/family and friend use, the resource inefficiency is not that big a deal.