• m-p{3}@lemmy.ca
    link
    fedilink
    arrow-up
    18
    ·
    2 days ago

    Some instances like lemmy.world values stability over bleeding edge, and 0.19.3 has been stable for them. They do plan to upgrade to something newer eventually, it’s just that every migration carries a risk of something going wrong.

    • gigachad@sh.itjust.works
      link
      fedilink
      arrow-up
      5
      ·
      2 days ago

      I remember we had some version a year ago (was it 0.18?) that made the whole lemmyverse unreliable. Federation was completely fucked, inage upload was buggy… some instances needed weeks for repairing the damage.

  • Lasherz@lemmy.world
    link
    fedilink
    arrow-up
    10
    ·
    edit-2
    2 days ago

    I imagine it’s similar to an enterprise not using Windows update. If it’s stable and no exploits have been identified then why rock the boat?

    • Mee@reddthat.comOP
      link
      fedilink
      English
      arrow-up
      1
      ·
      2 days ago

      On windows you got a enterprise or LTS equivalent to get the important updates.

      You don’t get that on Lemmy.

  • tal@lemmy.today
    link
    fedilink
    English
    arrow-up
    3
    ·
    edit-2
    2 days ago

    In the past, there have been some pretty unpleasant regressions.

    My own home instance, lemmy.today, had some time where it was more-or-less unusable, because every release for a while had some new regression. The lemmy.world guys were a lot more conservative, just backported some critical fixes and waited for a while after each new release to wait and see if problems showed up. They didn’t crash into the regressions.

    Granted, some of this could probably be picked up by better automated testing. But to some extent, I think that for at least big instances, it’s good to hold off, wait, and see if a new release has a bunch of issues.

    Also, my understanding is that at least for some (all?) past updates, there’s no downgrade path. Once you upgrade, you’re committed.

    Maybe you can back up the instances and restore them, but I suspect that that may break state across instances, since you’d get instances with conflicting views of what’s on an instance.