Jim Huang

The Linux Kernel Module Programming Guide (#LKMPG) got 5000+ stars on GitHub. Thank all the contributors and promoters. We are moving forward to Linux v6.x support. Rust LKM examples will appear later. github.com/sysprog21/lkmpg

Rob Pike's 5 Rules of Programming "write stupid code that uses smart objects" is my favorite one! cs.unc.edu/~stotts/COMP59…

求勝心會帶人提昇，達到更美、更善、更有用的境界。但怕輸是不同的心態，帶來的是停滯，因為不作事就不會因為自己的行為而失敗。 熱力學 (thermodynamics) 用「死寂的宇宙」 (heat death of the universe) 來描述熵 (entropy) 達到最高值的封閉系統 (closed system)。此時能量完全均衡，系統完全停滯，不再有變化。 求勝是德性，但怕輸不是。

Jiasin Yu： 藍白兩黨「國民戰隊」成形，第一砲公布協議就是個大笑話。 前面說「應與母體無顯著差異」，後面又說p-value應小於0.05。 欸欸欸！黃國昌主席不是說他做司法統計的嗎？p值小於0.05不就是「統計上的顯著差異」的意思嗎？！📷 然後所有民調界的前輩學長姐們，你們哪個敢保證未加權前樣本的性別、年齡、教育程度和行政區可以直接過代表性檢定？ 哭枵哩！我未加權就能過代表性檢定的話，我幹嘛加權啊！📷 未加權的市話民調，性別分布也許可以勉強通過代表性檢定。但年齡和教育程度分布要過檢定有多難知道嗎？ 如果用隨機抽樣訪問，就算加上戶中抽樣加重年輕人比例，年齡都不一定能過檢定，而教育程度分布幾乎不可能達到。更何況還有行政區。 要達成的可能性其一是「配額」。但四個變數同時配額，這難度有點高。且配額有配額的問題，你們又沒說清楚。 至於另一個難度更高，就是動員守電話的選民「造假」，例如都說自己20歲，但這樣反而有可能弄巧成拙。而且動員強度拉高，實際上對樣本的影響可能也有限，成本上划不來。 總之，這就是一份絕對合不了的藍白合協議，寫出這些內容的人，如果不是統計學的大外行，就是下定決心搞破壞的專家。 但會寫出p值小於0.05還要無顯著差異，怎麼想都是大外行。

Dennis Ritchie created C in the early 1970s without Google, Stack Overflow, GitHub, or any AI ( Claude, Cursor, Codex) assistant. - No VC funding. - No viral launch. - No TED talk. - Just two engineers at Bell Labs. A terminal. And a problem to solve. He built a language that fit in kilobytes. 50 years later, it runs everything. Linux kernel. Windows. macOS. Every iPhone. Every Android. NASA’s deep space probes. The International Space Station. > Python borrowed from it. > Java borrowed from it. > JavaScript borrowed from it. If you have ever written a single line of code in any language, you did it in Dennis Ritchie’s shadow. He died in 2011. The same week as Steve Jobs. Jobs got the front pages. Ritchie got silence. This Legend deserves to be celebrated.

Announcing rlsw, the new raylib software renderer backend. No GPU required. The Future is Now. #raylib

#zhtw-mcp helps AI systems produce Traditional Chinese aligned with Taiwan’s usage. It detects typical AI-generated writing patterns and rewrites them into more natural human expressions. github.com/sysprog21/zhtw…

Different Thinking of Genius.

just picked up this bad boy. can't wait to write some software with it

In 1887, Heinrich Hertz noticed that some light made it easier for electrons to escape a metal surface, but the pattern made no sense under Classical Physics. The surprise was that very bright red light could fail, while dim blue or ultraviolet light worked immediately. Einstein explained this in 1905 by proposing that light comes in energy packets, and only packets with enough energy can knock electrons free. That was the Photoelectric Effect, and it helped earn him the 1921 Nobel Prize in Physics.

#zhtw-mcp is a linguistic linter for Traditional Chinese that enforces Ministry of Education standards on vocabulary, punctuation, and character shapes, plugging through MCP and catching Mainland Chinese regional drift before it reaches the user. github.com/sysprog21/zhtw…

Major breakthrough tonight! After several weeks getting the C-based Prism Ruby parser hooked up to the Chicory WASM runtime (AOT compiled to JVM bytecode), I now have a local version of the gem that uses *zero* native libraries. Write once run anywhere indeed!

@BoWang87 Knuth naming it "Claude's Cycles" is the academic equivalent of a standing ovation. The man who spent decades hand-crafting algorithms just watched one get cracked in an hour by pattern recognition he couldn't replicate.

Donald Knuth is vibemathing now. real tough day for the stochastic-parrot crew.

g ≈ π² is not a coincidence. The original idea for the definition of “1 meter” was the length of a pendulum with a half-period of 1 second. The formula for a pendulum’s period is: T = 2π √(L / g) If you plug in L = 1 m and T = 2 s, you get g = π² m/s². The actual definition of “1 meter” was eventually chosen as 10⁻⁷ of a half-meridian passing through Paris, but this was inspired by the earlier pendulum-based definition. The two definitions agree to within 1%.

Windows 11 still has icons for Borland Turbo Pascal and WordPerfect.

The Claude C Compiler is the first AI-generated compiler that builds complex C code, built by @AnthropicAI. Reactions ranged from dismissal as "AI nonsense" to "SW is over": both takes miss the point. As a compiler🐉 expert and experienced SW leader, I see a lot to learn: 👇

We just released a new version of Kitten TTS - 15M param SOTA tiny text-to-speech model It has a significant quality improvement over the previous version. Still less than 25MB in size! Open-source, extremely tiny, expressive. github.com/KittenML/Kitte… Apache 2.0

3,488 triangle edition of the Utah teapot, environment mapped by a hardware geometry engine through my recreation of the SST-1 fixed function pipeline. Important note: because it is the Utah teapot, the bottom is missing. Wouldn't be right to add it. 100% FPGA hardware. Entirely SystemVerilog.

"Inside every large program there is a small program trying to get out." -- Sir Tony Hoare

In today’s daily dose of hand-written SH4 assembly for the Sega Dreamcast, I’m attempting to implement a glOrtho()-equivalent routine within my fast math and linear algebra library, SH4ZAM, to either initialize the current matrix to be a 2D orthographic projection matrix, or apply it onto the current matrix. I offer at VERY LEAST these two forms of GL-style transform functions, typically offering even more variants. Why? Cuz ain’t nobody got time to initialize the matrix to identity just to multiply it back by another matrix just to arrive at something you could’ve directly initialized the current matrix to, saving cycles. Thinking only in terms of GL-style full 4x4 matrix multiplication for every transform is a fantastic way to waste clock cycles… I’ve written a small example of how to be more optimal with my API here: #autotoc_md60" target="_blank" rel="nofollow noopener">sh4zam.falcogirgis.net/guide.html#aut…