Skip Navigation

Posts
2
Comments
318
Joined
3 yr. ago

  • awk is in a strange position between simple grep commands and full-featured Python programs. It's powerful but too domain-specific, and if you learned how to write multiline sed scripts you don't need awk anymore.

    And yes, you can do

    python -e "..."

    for (really long) Python one-liners.

  • Tech companies spend effort on a FOSS project when either it's their main product, or when they have no choice, it's licensed under GPL and there are no BSD or Apache-licensed alternatives. Contributions are usually done by individual employees in their after-hours time, and most managers see it as directly benefitting their competition.

  • Other companies provide nothing, so it's definitely better than average.

  • Nah, the kids I know learned C just fine. C is simple, C is all you need for writing kernel drivers, pointers are not that hard if you explain them well, it just feels really pedestrian compared to Python or Typescript.

  • It's ultimately a question of money. Older guys with software engineering degrees and fancy salaries can spend their weekends doing free community service in the form of open-source development. Younger people have to worry about job and rent and bills, they simply don't have that kind of free time.

    Add to that the growing complexity of the software. Something that could be done by an university student before, like writing an OS from scratch, won't be nearly as useful as it would in the '90-s, because it was already done before, now you have multiple OSes to choose from. And joining an existing software project is hit-or-miss, some are inclusive and some are an old boy club where you need to know the secret rules.

  • It's not for pilot holes, it's for situations when you strip your screw head but continue screwing the hole out of spite.

  • In Lua all arrays are just dictionaries with integer keys, a[0] will work just fine. It's just that all built-in functions will expect arrays that start with index 1.

  • You can actually use / as a path separator on Windows in functions like fopen(), because it supports some ancient version of POSIX standard.

  • Missing a Raspberry PI 4 setup which hosts a print server, an RTP server with two surveillance webcams and no password, and also seeds a terabyte of torrents over the local flower shop's unencrypted WiFi.

  • Oh the heavens above, it supports touch controls! It's the first 'hack' port of a game with mobile support I've see.

  • There's nothing wrong with kilowatts, it's an SI unit. The problem is hour, which is 3600 seconds, and we have ancient Egyptians to blame for this, who divided the day into 24 hours despite having already developed base-10 numerical system.

    Kilowatt per kilosecond, which is 1 megajoule, would work better.

  • what made it feasible in 2025 that changed since 1950.

    It's obviously all that aerodynamics simulation and wing shape optimization that we can do using modern computing power. Those airflow equations were a bitch to calculate in 1950, and that did not change in 2025, but you can now throw them all onto your GPU and make it loop theough various body shapes to find one with most lift under every possible air speed and angle, and congratulations, you've created an AI to design planes.

    I don't think we'll see civil airliners designed that way, the tubular body is the structurally strongest shape for the amount of metal it uses, and the wing shape is already optimized to maximum. I bet the full-sized BWB plane will use some fancy titanium alloy that will make it too expensive for non-military to produce.

  • It's not particularly harder, one year manufacturing time for a full-size prototype airplane is kinda fast.

    It's also very different from 80-years-old flying wing design, that time it was literally a wing with no body, and had problems with pitching and turning, this time it's more like a body acts like an additional wing.

  • Jumping from loose-typed language to strict-typed language will be hard.

    It's also a matter of your general programming experience. Once you write, like, ten thousand lines of meaningful code in Python, learning C# should take you a month or two at most, you'll know most programming concepts and algorithms intrinsically, and the rest is just learning syntax.

  • I don't think I care about shape, as long as it's made from durum wheat. Now, we have a lot of pasta here that's made from regular baking flour, it's still very common in EE countries, and it's damn cheap. You must boil it for 40 seconds and not a second more, or it instantly clumps all together and turns into a wallpaper glue.

  • Desperately needs strafe on-screen keys

  • It makes sense, digestive system is way better at filtering harmful chemicals than lungs.

  • It's just C++ without templates. So whatever element you want to put into std::list, it must virtually inherit std::list::value_type. And of course there will be a macro-based list monstrosity inside Boost.

  • So is it 'QR-style codes' or just QR codes? What's the encoding - QR code, AZTEC, Data Matrix, or PDF 417?

  • AI is currently all the rage. I know a guy who does neural networks training, and it's the closest job to Pokemon trainer that you can have. He does not know any programming language, it's all purely data processing, and it requires a heavy math background. It's still an 8 hours per day job in a corporate environment, and a half of it is fixing his computing farm, because there's no dedicated sysadmin for replacing burned videocards.

    Another math-heavy option is signal processing. I've worked in a company which produces wireless communication hardware, it's like 200 people, and there were two signal processing guys who patented an algorithm which increases data throughput by something like 0.5% in some specific high-noise scenarios, they are listed together with CEO on their website About page. That said, all the easier tasks for signal processing are already done, we're already at the theoretical limit of Shannon theorem, so you'll only be making gradual improvements.

    I would also recommend cryptography research, but it requires a lot of coding knowledge, even if you're not writing any actual code, and paranoia is pretty much a requirement.