Comments by "Michael Lenczewski" (@kayakMike1000) on "Low Level" channel.

  1.  @bozingo  functional code NOT horrible. 😂 😂. Ha ha ha ha ..
    1
  2. Alternatively, you could learn how to write C well enough so you don't corrupt memory. Static analysis could help establish enough evidence for memory safety.
    1
  3. It's not so much that rust runs on everything, you just have a toolchain that compiles to whatever janky BS that is AVR.
    1
  4. Plain C is supposed to be snake case!
    1
  5. No. All you need is C.
    1
  6. ​ @everyhandletaken  ha ha oxidation... Rust. Well played sir
    1
  7. Rust still sucks. What does the whitehouse actually know?
    1
  8. Why can't the ideas in rust be implemented for C through some sort of linting or other static analysis?
    1
  9. ØMQ is a facade then? It kinda sounds like a Plan9 channel.
    1
  10. Does this trigger a hard fault like IRQ 3 in arm cortex-m0+?
    1
  11. Well... Technically... Interpretation vs decoding in the CPU architecture is a wee bit of a gray area. You could design any architecture to be this way...
    1
  12. No... No... Do fixed point math. Edit: Oh yeah man. You got it man. Godbolt FTW.
    1
  13. Hardware makes randomness pretty good. Need hardware module to measure entropy in the environment. Hmm.... What if you could get entropy based on the weather report, direction or the wind, or I dunno just put an antenna that listens to static?
    1
  14. ​ @y00t00b3r  400 pounds of prefrontal cortex. I am just a brain in a jar of nutrient solution.
    1
  15. Ew. I think i will just use better pointer hygiene
    1
  16. Magic number, the number 42 is. Linting your code you shall, errors avoid you will.
    1
  17. Also... Recursive functions are not really deterministic, which is probably something important in an embedded system, you often want predictable behavior
    1
  18. F
    1
  19. How dare you make this so much simpler! You must maintain the dark arts of firmware!!!
    1
  20. Magic numbers inline bad. Use macro.
    1
  21. Oh. I watched this already....
    1
  22. Hum... I am working on the same thing, though I am targeting a few different chips... Sorta built-in portability. So... Lemme get your opinion. The OS needs to set up each process stack (and initial context frame), each process heap, schedule each process, and provide mechanism for IPC?
    1
  23. CMake is the devil incarnate! Meson is better.
    1
  24. F
    1
  25. First transistors were made of germanium, not silicon.
    1
  26. Rust is silly.
    1
  27.  @9SMTM6  well sure, FETs are more fundamental than CMOS, but the complimentary part is an essential to the design. In most logical like a flip flop, there's equal numbers of PMOS and NMOS components. I think there's at least 4 or 5 pairs in a CMOS flipflop for a total of 8 or 10 transistors per bit. Wikipedia is cool, but take it with a pound of salt. The complimentary part is important to make these "gates" reliable, energy efficient, switch fast, avoid latching, etc. etc. CMOS is the best refactor optimized for binary operations. It's way more than just a manufacturing technique. But technically, you're correct. It's more like FET is important.
    1
  28. Look up tables? Edit. Nope... I was wrong...
    1
  29. Yup. I learned the same thing in my physics degree. Experience matters more.
    1
  30. These restrictions are actually pretty durn good for embedded programing.
    1
  31. ​@William Drum infinite recursion loops are a problem. I recall an XML parser that would run into memory issues loading an XML with a billion lols. I think that was a heap issue, not a stack smashing issue, but it was an example of recursion being somewhT bad.
    1
  32. I brought this in for show and tell at work... They adopted it for a piece of legacy firmware.
    1
  33. Parallelllilize.
    1
  34. X86 isn't arm.
    1
  35. Design your own cpu based on the riscv architecture
    1
  36. Spinnaker has thousands of ARM968E custom CPU cores. I think each has a bank of tightly coupled memory for each core.
    1
  37. Magic numbers kinda suck too, but that's a mistake in all languages. Also... What's up with that noop? Are there not proper timers? Also ... Screw make. Get Meson and Ninja. Much much better
    1
  38. Linux is a great application that runs pretty well on my windows PC.
    1
  39. Hmm... Unlikely encrypted. They licenced an arm core for their ASIC. There's a possibility they implemented encryption in hardware.
    1
  40. Someone recently found a alpha64 windows2000 install. It was never released, but they wrote it. They even found a compiler forit.
    1
  41. Why? They are just null terminated sequences of bytes. It's a fairly easy data structure, one of the easiest.
    1
  42. Now do OPAQUE pointers!
    1