Menu ⇓

  2. Astonishing UI design at the OBI self-checkout today. For non-german-speakers, the text reads “Please think about the heyOBI App!” and the “answer options” require no translation

    (the two answer options read “1. ok” and “2. ok”. You can also choose to cancel, which is apparenly something different?

  3. Small PSA for anyone else late to the game: don’t bother buying Elden Ring on the bandai namco store, it’s just a slow, inconvenient way of getting a steam code via email.

  7. In reply to a post on twitter.com

    @seaotta the best way is to start writing them!

    But IMO a big part is normalising the idea that anything which people would currently write as a tweet thread would be better off as a blog post. Normalising the idea the blog posts don’t need to be polished, or to have stock photos, or even a title.

    And then build on that: tweet what would have been the first tweet of the thread, with a link to the full post. Use something like brid.gy to gather responses from twitter as a comment thread.

    Might be worth replying to threads we enjoy with “hey, I love your writing, do you have a blog?” to remind people that they exist and would be read

  8. ↪ Aaron Parecki: I sold a washer/dryer on Facebook Marketplace, and now it's showing me new listings for washer/dryers. Glad to see the algorithm hard at work here.

    Apparently FB learned nothing from the classic “I bought a toilet seat on amazon and now amazon’s trying as hard as it can to fuel the toilet-seat-collecting hobby I’m apparently hooked on” problem

  9. A little-appreciated benefit of (British) English speakers learning a european language: you can look up recipes in that language and be guaranteed that they’ll use metric units

  10. This amazing comic from @Gingerhazing proves yet again that the best thing to come out of Star Wars is the lovingly crafted fan works which go places and tell stories the originals never would have dared.

    Parts 1-2 https://twitter.com/Gingerhazing/status/1489747577541070850
    Parts 3-8 https://twitter.com/Gingerhazing/status/1491149997676822530
    Epilogue https://twitter.com/Gingerhazing/status/1494087642081742849

    (Really hoping that they compile these into a nice readable format somewhere, it is just too good to be lost in the depths of twitter)

    Edit: there’s a full version up on substack!

  12. Annoyed by websites hijacking your favourite browser keyboard shortcuts? Here’s how to disable it in firefox (tested in ff 95, probably works in other versions).

    On a site-by-site basis: Go Tools → Page Info (cmd/ctrl + I also works, if it’s not hijacked), and block keyboard shortcuts:

    Or you can disable it browser-wide so that it never bothers you again, in about:config, by setting permissions.default.shortcuts to 2.

    In theory, 3 should prompt you on a site-by-site basis, but it doesn’t seem to work, sadly.

  15. I thought I was the only person who celebrated the spider of the year, but I heard that NASA just sent a big web into space?? Cool of them to get in the festive spirit like that

  17. Is there an institution or organisation somewhere whose explicit goal is to constantly run unit tests on basic physical phenomena, just in case, say, the gravitational constant suddenly changes or something?

  19. It appears that the popular Riden/Ruideng/RD Tech DP/DPS/DPH series power supplies use low-side current sensing, which can lead to some unexpected (and potentially destructive) behaviour in a situation where you have multiple unisolated power rails.

    For example, I put together a little box with a DP30V5A providing a variable, current-limited rail, and three LM2595 modules providing fixed 12v, 5v and 3.3v rails. As none of these supplies are isolated, and can therefore not be used to provide negative rails, I tied all their 0V outputs together. This led to the DP30V5A reporting a completely false current consumption of about 33% of the measured value.

    In the test setup, the DP30V5A is set to source 10V at a maximum of 5A. It’s connected (via a multimeter in 10A current measurement mode) to an electronic load set to consume 1A. Both the multimeter and load report a current consumption of 1A, but the DP30V5A reports only 0.29A.

    After sketching everything out, it became obvious that this was due to the low-side current sense resistor only seeing some of the current flow, and the rest flowing through the unused LM2596 modules (the switch A represents the internal connection between the DP30V5A 0V and the fixed rail 0V)

    Disconnecting the 0V rails and providing a separate 0V binding point for the fixed rails fixed this issue, and I’ll just have to keep in mind that if I want to use multiple rails from this mini PSU in the same circuit, I can’t trust the DP30V5A current reading and have to set its maximum current to about 33% of the desired value. Otherwise, the software overcurrent protection can’t function correctly, and there’s a risk of damaging both the module and the circut under test.

    An amusing side effect of this setup is that the DP30V5A low side current sensing can be used for the fixed rails! I doubt I’ll ever encounter a situation where this is useful though.

    More about high and low side current sensing in this AAC article