I was wondering whether that could cause issues for the limited servers available, like when the mass migration from reddit happened. Would it be bad to add more accounts that would be mostly dormant? And if one does create a throwaway would it be better to delete the account afterwards.

  • fiat_lux@kbin.social
    link
    fedilink
    arrow-up
    23
    ·
    edit-2
    10 months ago

    Disclaimer: I am not a Lemmy dev.

    It would add some extra rows to a database, which can increase lookup times if enough people do it with enough accounts… so from a general engineering perspective, I wouldn’t encourage it.

    In a more realistic sense, it would take a lot of people and/or a lot of throwaways to effect much difference. That is, assuming the database queries aren’t too complex or inefficient, and the servers aren’t nearing critical capacity.

    Deleting the account afterwards may not be as effective as never creating the account altogether. There’s a chance some stuff is only tombstoned instead of deleted, things get stuck in caches too, it would probably be better than keeping the dormant accounts though.

    Tl;dr It would be polite of you to keep only a couple of throwaway accounts, but I wouldn’t feel too guilty about making them. Just don’t be like a spam bot and create dozens or more.

    • ursakhiin@beehaw.org
      link
      fedilink
      arrow-up
      8
      ·
      edit-2
      10 months ago

      Depending on implementation, the lookup should be indexed so the time difference would be very minimal on even large tables.

      The cutest of storage could be a problem depending on how many wasted accounts exist, but even that should require a ton of accounts to make an impact.

      • fiat_lux@kbin.social
        link
        fedilink
        arrow-up
        2
        ·
        edit-2
        10 months ago

        Yeah, there’s always room for a lot of implementation-dependent possibilities of good DB practices, but it’s also possible for one mistaken PR approval to create exponential load.

        In the case of FOSS where there might be fewer formal processes to catch errors, I like to err more heavily on the side of caution. If for no other reason than preventing surprise server scaling bills for the volunteer admins.

        Also: if you are making throwaways, please consider donating time and/or cash to those instances.