I met Steve Bourne a while back and asked him why he chose to write sh with all those ALGOL-like macros. His answer was terse—he just wanted to use the syntax he was most familiar with.
Busybox is where I learned the multi-call technique of using argv[0] to simulate multiple commands in a single binary.
It works great for Rust binaries, which can get quite large. I had to rewrite some Python utilities (software updater, Wi-Fi network config, etc.) for a very constrained embedded system (64MB RAM) that couldn’t handle Python. So I built them all in a single multi-call binary. I believe `clap` has built-in support for it.
Also, Busybox has a lot of stuff built in that you’d never think of. It has a little vi built in. It has the full suite of ‘runit’ process supervisor utilities. And plenty more. It has a Kconfig system (like the kernel) where you can tune exactly which utilities you need and what features you need. Don’t need full output in ps? Turn it off. Don’t need tab completion? Pretty sure you can turn that off too. Always typing ‘dc’ instead of ‘cd’? Just don’t include dc.
This was a class assignment in the 15-213 class at Carnegie Mellon. The staff had set up a test suite and an online leaderboard to rank the speed of each student's malloc implementation.
I figured out that the test cases allocated a disproportionate amount of X-byte blocks. I was able to get to the top by hardcoding a specific freelist just for X-byte blocks.
Learned a lesson about easily it is to game a benchmark :)
This is a useful lesson! Tuning an allocator for a single application reduces to optimizing for that application’s empirical allocation patterns (sizes, lifetimes, access, usage).
I think this is me. I was diagnosed with “regular” ADHD a few years ago, but I’ve had issues with rejection sensitive dysphoria my whole life. As a kid I would have a lot of meltdowns when I couldn’t get something right on the first try or made any kind of mistake.
The meltdowns stopped, but I still have issues spiralling into thoughts of failure and being a horrible person when I feel like I’ve disappointed family or friends.
Guanfacine has helped though. Tried a bunch of medications and this is the only one that seems to have made an impact.
Hey, you've basically described me to a tee as a boy. I had the exact same temperament: meltdowns when I thought I disappointed people, putting myself down all the time for failures and mistakes. My parents weren't even particularly strict.
Might be worth giving Qelbree a shot as well. I switched to it last year and...yeah, it' was a complete game changer. Hard to get insurance to cover, but it's been well worth it.
On early bit-paired keyboards with parallel 7-bit outputs, possibly going back to mechanical teletypes, I think holding Control literally tied the upper two bits to zero. (citation needed)
Also explains why there is no difference between Ctrl-x and Ctrl-Shift-x.
Battlezone and later 3D games used different circuitry that generated vectors with analog circuitry and used a “math box” coprocessor based on Am2901 bit slice ALUs to help with the 3D math.
Growing up in New England at the time, I was very lucky that they ran a pilot project at my elementary school. We programmed our "sketches" (though we didn't use this term) in LogoWriter on Mac LC IIs and flashed them using a weird dongle that connected to the brick via RJ11 cable.
I owe a lot to my teachers and Fred Martin's group. I don't think I'd be a developer today without that experience.
I was fortunate enough to get LEGO Mindstorms (the OG one) for Christmas as a kid. By that time I'd already learned a bit of QBasic and so had programming experience. While I found the visual programming paradigm limiting I also wasn't creative enough to come up with robots that needed more than what one could build with the software so I had no complaints. I had a great time learning about gear ratios and how flaky the light sensor was depending upon room brightness.
Is there anything similar and as well-supported from Lego or otherwise these days? An ESP32 is great fun but some assembly is required.
I was very surprised that GCC could optimize NEON SIMD intrinsics. After spending hours trying to optimize my vector code, trying to get the spacing between register dependencies right to reduce stalls, breaking long reduction operations into intermediate results, messing with LLVM-MCA, etc., I realized that I just couldn’t beat the compiler. It was doing its best to allocate registers and reorder instructions to keep the pipeline filled.
I don’t think it always did the best job and saw a bunch of register spills I thought were unnecessary, but I couldn’t justify the time and effort to do it in assembly…
Storage tubes are amazing. I once had a Tektronix 464 scope. Unlike the ubiquitous 465, the 464 was an analog storage oscilloscope! The persistence of the storage tube let you capture pulses, visualize signal jitter, etc. And there was a very satisfying erase button that cleared the tube with the characteristic green flash.
You could also just mess with the horizontal/vertical position knobs and use it as a very expensive Etch-a-Sketch.
Unlike a television CRT, the storage tube will "remember" everywhere the electron beam struck the screen as long as the voltage is below the erase threshold.
What's wild is that this information can be read back out of the device. So, it's not just a display but also memory. It's a destructive reading process (not unlike core memory) where detection is also erasure, but it really is RAM... of a sort.
