• pHr34kY@lemmy.world
      link
      fedilink
      arrow-up
      3
      ·
      16 days ago

      I recently made the jump. It’s crazy how much effort goes into optimising Nextcloud it still remains the most demanding thing on your server and your sanity.

      • mholiv@lemmy.world
        link
        fedilink
        arrow-up
        1
        ·
        16 days ago

        Hot take. Nextcloud is hot garbage. If you want file sync use Seafile. So so so so much faster and more resource efficient. For the rest of next cloud functionality just use a tool specific to that.

        Nextcloud is still tied to the old php/fpm model. There are better approaches now days. Hell even owncloud is rewriting away from the php/fpm approach.

        • Echedelle (she/her)@lemmy.blahaj.zone
          link
          fedilink
          arrow-up
          0
          ·
          15 days ago

          The problem is not php-fpm, the problem is when people who perform as SysAdmin dont know how to tweak it properly.

          I deploy Nextcloud instances for production environments. :/

          • Taasz/Woof@piefed.social
            link
            fedilink
            English
            arrow-up
            1
            ·
            14 days ago

            Why would I need to tweak it when what I ran was provided by Nextcloud as a complete package?

            I’ve also never used a commercial Nextcloud host that has acceptable performance.

            If they are providing their AIO Nextcloud unoptimized that’s even more of a reason to dislike it.

              • Taasz/Woof@piefed.social
                link
                fedilink
                English
                arrow-up
                1
                ·
                13 days ago

                More just saying that I’ve never had a nextcloud instance perform well, regardless of where or how it’s hosted.

      • Jason2357@lemmy.ca
        link
        fedilink
        arrow-up
        0
        ·
        16 days ago

        I am just about to start experimenting with opencloud.eu. It is another developer revolt fork of Opencloud (this time their infinite scale project) so it is supposed to be a lightweight single binary that uses a real filesystem and no database at all. Lacks all the add ons of nextcloud, buy I am tired of maintaining it.

        Anyone try it?

        • myszka@lemmy.ml
          link
          fedilink
          arrow-up
          1
          ·
          15 days ago

          uses a real filesystem and no database at all

          Sounds very promising! That’s what has always pushed me off Nextcloud.

        • T4V0@lemmy.world
          link
          fedilink
          arrow-up
          1
          ·
          15 days ago

          Yep, though I have a bug in my use case from a older version, and I still didn’t have the time to try the new stable version. I also translated the Docker compose for the project and extensions into .container files. In the newer version it seems they managed to remove a sidecar container.

          So far, it runs on a VPS with a 4-core and 8GB of RAM. On a first run, after upgrading, it needed swap as the system started freezing and the OOM killer stopped the program, zram was enough in my case.

          Besides the main container — and the sidecar — I used the Collabora container, LDAP for user management, Keycloak for auth, full Tika for OCR search — modified for my language — and Radicale. Mind you, most probably want the main container and the online office container, be it EuroOffice, Collabora or OnlyOffice.

          Here is the memory usage and CPU usage on standby from podman stats:

          NAME                       MEM USAGE / LIMIT  CPU %
          systemd-tika               175MB / 8.054GB    0.68%
          caddy                      43.35MB / 8.054GB  0.24%
          systemd-ldap               67.08MB / 8.054GB  0.01%
          radicale                   2.425MB / 8.054GB  0.00%
          syncthing                  29.97MB / 8.054GB  0.21%
          systemd-keycloak-postgres  41.12MB / 8.054GB  0.06%
          systemd-keycloak           663.4MB / 8.054GB  0.75%
          systemd-collabora          42.02MB / 8.054GB  0.27%
          systemd-opencloud          883.4MB / 8.054GB  7.31%
          systemd-collaboration      57.71MB / 8.054GB  0.31%
          
        • JustEnoughDucks@feddit.nl
          link
          fedilink
          arrow-up
          1
          ·
          15 days ago

          Setup is incredibly bad. They have a very very strict set of parameters to be able to log in at all. I have pretty much given up OK it as there is no way for me to log in for the first time with default admin credentials. It simply doesn’t accept the states environment variables on docker and you get 1 attempt to log in before it gets stuck in a loop and you have to purge the image and every volume and file to get a second attempt.

    • blinfabian@feddit.nlOP
      link
      fedilink
      English
      arrow-up
      0
      ·
      16 days ago

      tried it, but i couldnt figure out how to add my existing photos in the way i had organised them: /year/event/photo.jpeg and nextcloud made it pretty easy

      • Taasz/Woof@piefed.social
        link
        fedilink
        English
        arrow-up
        1
        ·
        14 days ago

        Hmm the exif data for date taken should organize them chronologically without needing the folders to import.

  • AceFuzzLord@lemmy.zip
    link
    fedilink
    arrow-up
    2
    ·
    16 days ago

    For me, I would personally make the adjustment:

    chatgpt —> NOTHING

    I’d rather go my whole life with people ridiculing me for not using genAI then hand over my status as a person to any form of genAI.

    • blinfabian@feddit.nlOP
      link
      fedilink
      English
      arrow-up
      1
      ·
      15 days ago

      yeah, imo search engines have gone completely to shit. and sometimes AI is literally the only way to find certain things without wasting a lot of time. but i do try to use it as little as possible

    • Zephyr@sh.itjust.works
      link
      fedilink
      arrow-up
      0
      ·
      16 days ago

      Do you feel that way about the other algorithms you do use? Nearly all code we do use was compiled and the compiler makes it’s own decisions on how to convert abstract code into assembly or binary. Do you have anything against machines making decisions on that conversion?

      • Auli@lemmy.ca
        link
        fedilink
        English
        arrow-up
        0
        ·
        15 days ago

        But that is not the same. I can compile the same code with the same compiler and get the exact same code. This is not true of AI.

        • Zephyr@sh.itjust.works
          link
          fedilink
          arrow-up
          0
          ·
          15 days ago

          I have a feeling you didn’t understand the question I was asking. Ilms are just one type of algorithm and data structure. There’s other algorithms nearly all programers use to compile symbolic code Into assembly or machine code (binary), they’re called compilers. Anyone who isn’t writing assembly or machine code directly is likely using a compiler.

          • pineapple@lemmy.ml
            link
            fedilink
            English
            arrow-up
            0
            ·
            15 days ago

            Your question is confusing to me, I know what compilers are but I’m not sure how that is super related to llm’s. I was just saying that people use llm’s for reasons aside from generating code.

            ooh wait nevermind, I realise what you meant now. Your comparing compilers to llm’s since they both are algorithms that turn an input into an output. Is this what you meant?

            • Zephyr@sh.itjust.works
              link
              fedilink
              arrow-up
              0
              ·
              edit-2
              15 days ago

              I got the comparison from Linus Torvalds talk but yes and depending on the compiler and it’s use the output might not be so deterministic. More accurate to the point I’m actually making is is that these are all algorithms and the issues people have with AI isn’t anything about the algorithms or the technology itself but the ways in which some people and companies are using it. May as well say you hate the internet because some people and some companies are using it to do very harmful things.

              Here’s another

              • pineapple@lemmy.ml
                link
                fedilink
                English
                arrow-up
                1
                ·
                14 days ago

                Interesting, I can see how the idea that llm’s are just another layer of abstraction like compilers. Although this doesn’t really justify using them and that unfunny joke of “I bet your code was written by a compiler” just doesn’t really make sense. It’s not like we should start writing all our code in python, for example.

                • Zephyr@sh.itjust.works
                  link
                  fedilink
                  arrow-up
                  1
                  ·
                  14 days ago

                  What would justify using them? For many I talk to they have an irrational hatred because of how some people and companies have used them so now it’s like a prejudice. That is not matter what they dislike neural networks

      • AceFuzzLord@lemmy.zip
        link
        fedilink
        arrow-up
        0
        ·
        15 days ago

        Compilers ain’t a problem for me. Compilers aren’t enabling people to claim writing they never wrote as their own so they can sell “books”. Compliers aren’t generating false information on purpose in order to trick me because it was built to be a gaslighting yes man¹. Compilers aren’t deleting anything from certain files to whole directories¹ off my system unless I tell them to.

        I don’t have to worry about a compiler trying to steal all my data¹ in order to please braindead C-suites trying to replace all humans with clankers. So I have zero issues with a good genAI free compiler.

        ¹ unless they have been compromised/my system has been infected

        • Zephyr@sh.itjust.works
          link
          fedilink
          arrow-up
          0
          ·
          15 days ago

          I mean it does allow people to claim code they didn’t necessarily write is there’s. It seems that your issues aren’t to do with LLMs fundamentally but specific use cases some individuals have chosen. Like you don’t fundamentally have an issue with the algorithm itself.

          • AceFuzzLord@lemmy.zip
            link
            fedilink
            arrow-up
            0
            ·
            15 days ago

            I will concede on the algorithm. Can’t get too mad at advanced maths, but I can get mad at the lazy people using it for nefarious purposes or just because they’re lazy.

            • Zephyr@sh.itjust.works
              link
              fedilink
              arrow-up
              0
              ·
              15 days ago

              I mean may as well say you’re mad at the internet because people use it for nefarious purposes as well. This is usually what I find about all these debates around AI and LLMs or NNs. It’s more so people being legitimately upset at actions taken by some individuals or companies more so than any hatred for the technology itself but they conflate the two effortlessly. Like we’re aware of how strange it sounds in different cases when that’s done but for some reason with AI most don’t see this issue in their argument.

          • Auli@lemmy.ca
            link
            fedilink
            English
            arrow-up
            0
            ·
            15 days ago

            What the hell are you talking about. Compilers do not insert code that wasn’t put there. Now if your saying well you wrote C and the machine is running assembler then you are a moron.

            • Zephyr@sh.itjust.works
              link
              fedilink
              arrow-up
              1
              ·
              edit-2
              15 days ago

              If I’m a moron then so is Linus Torvalds since I literally stole the analogy from him. Generally speaking though these are all just algorithms and the issues people have aren’t about this particular algorithm but the ways in which some people or companies are using the algorithm. It’s a bit like saying you hate the internet because some people use the internet to do very harmful things.

              Here’s another

  • FeelThePower@lemmy.dbzer0.com
    link
    fedilink
    arrow-up
    0
    ·
    16 days ago

    unfortunately every foss maps app I’ve tried has been subpar. all the ones based on open street maps can’t even go to proper addresses, it takes you to the street and you just have to look around for it. at least here in the us. so far magic earth has been better but it still doesn’t work with Public transit so I have to have a separate maps app in my private space anyways.

  • KaChilde@sh.itjust.works
    link
    fedilink
    arrow-up
    0
    ·
    16 days ago

    My only major change would be to remove Mullvad.

    One of the founders has made major donations to far-right groups in Sweden. People have argues that it’s his money to donate, but if paying for Mullvad puts money into the pockets of a major donor to the far-right, I’d rather support an alternative provider.

    I have had a very similar user experience with IVPN, and I have heard people suggest AirVPN as a similar alternative.

    • JoyfulCodingGuy@lemmy.ml
      link
      fedilink
      arrow-up
      1
      ·
      16 days ago

      As someone else started it’s worth noting that Mullvad themselves did not donate to anything. I understand people’s argument that money to Mullvad goes to the founder and thus goes to the people they donate to and that was people’s argument for Proton as well. So do what you will as it’s everyone’s personal preference.

      That’s why lists like these, while useful for those who are just looking into privacy, deGoogle, etc. to first learn of projects and services, are almost always lacking nuance without text or all the major alternatives where then people can choose based on their beliefs. 🙂

      Note: I understand this is one person’s list and not a general list but just stating my thoughts on the matter as a whole.

        • Maudelix@lemmy.world
          link
          fedilink
          arrow-up
          0
          ·
          16 days ago

          I don’t see a difference honestly, plus you know there is underhand contributions going on.

      • poke@sh.itjust.works
        link
        fedilink
        arrow-up
        0
        ·
        16 days ago

        They haven’t donated to political parties, have they? That would have been big news I’m pretty sure I would have seen.

        • iamthetot@piefed.ca
          link
          fedilink
          English
          arrow-up
          1
          ·
          16 days ago

          Not aware of donations, but Proton’s CEO was glazing Trump around the start of his second term.

  • thatKamGuy@sh.itjust.works
    link
    fedilink
    arrow-up
    0
    ·
    15 days ago

    I honestly wish I could quit Excel - it’s literally the last vestige of Microsoft in my non- professional life, but I find LibreOffice Calc a bit too needlessly different to be able to fully switch across.

  • AnimalsDream@slrpnk.net
    link
    fedilink
    English
    arrow-up
    0
    ·
    15 days ago

    This setup is telling me you’re spending too much time on computers. Consider a hobby where you make something with your hands, using tools that aren’t electrified.