our version of lemmy is old enough that clients like mlem are starting to break due to API drift, so I’m finally upgrading us to the latest stable version of lemmy. this will involve a bit of downtime and potentially a number of breakages; keep an eye out for anything that doesn’t look right after the upgrade and let us know!

  • froztbyte@awful.systemsEnglish
    0·
    9 days ago

    for the viewers at home: you definitely don’t want to have to do this yourself

    (we’re going over the lemmy backend changelog and hooo boy)

    • 0_o7@lemmy.dbzer0.comEnglish
      0·
      9 days ago

      So you guys keep running old software till something breaks and are surprised new software that you haven’t kept yourself updated with, has breaking changes…?

      ohhh boyyyy

      • froztbyte@awful.systemsEnglish
        0·
        9 days ago

        thanks for your valuable and insightful contribution, I’m sure we would’ve been utterly at sea without it. simply marooned on the high seas of software, adrift and helpless. thank you, 0_o7, for your sage advice! you have saved us!

        wait, hang on. no. the other thing.

      • self@awful.systemsOPMEnglish
        0·
        9 days ago

        it’s weird you’re not posting this from your own lemmy instance running a bleeding edge version — you know, one of the 1.0.0 versions that don’t fucking work

      • froztbyte@awful.systemsEnglish
        0·
        9 days ago

        an absolute clowncar of yolo development practices. here’s just a few of the notes (there’s a whole spreadsheet):

        scripts that used to be called x.sh (shebang /bin/sh) suddenly renamed x.bash in main (shebang still /bin/sh). most of them take cli args, have no safety nets (arg checks etc), and have side effects (file alterations, commits and tags generated, etc)

        there’s a RELEASES.md which afaict is/was not generated by anything ever in the repo. it’s also out of date, has been touched by 5 commits (one more if you include other branches), and is wrong

        the build and test process targets both a rust stable version (1.81) and nightly

        the container compose files fully switched out from the mainline postgres containers, to something called pgautoupgrade. I haven’t yet gotten to look to see what it is/does, or what magic footguns it might be. the inter-version migration testing appears to be fully reliant on this thing.

        the sum test action for migrations is “run sed over a target config, then spin it up”

        there appear to be some new config settings. they have comments about their usage. they appear to be important. I only caught them because I was looking at the 0.19.3..0.19.12 diff

          • froztbyte@awful.systemsEnglish
            0·
            9 days ago

            what is the decision process there

            why think at all if you can just tread water and iterate 37 times instead?

            (I don’t really know what the thinking is, can only see the outcomes. but it’s some hella mediocre no-effort shit, punching through the problem)

        • flowerysong@awful.systemsEnglish
          0·
          9 days ago

          scripts that used to be called x.sh (shebang /bin/sh) suddenly renamed x.bash in main (shebang still /bin/sh).

          That’s just…so completely ass-backwards it boggles the mind.

        • froztbyte@awful.systemsEnglish
          0·
          9 days ago

          ah, cool. so, themes? the thing that almost every software out there makes as a lightweight pluggable?

          not lemmy, no. in lemmy you recompile your app to get 'em. (source)

          (edit: there’s a special kind of thinking that goes into writing code this manner. this is pluggable (see extra_themes_dir) but consider the thinking that goes into writing code like this. “some bits static”, with a maintenance burden and a recompile overhead. it’s just absolutely fucking dumb)