<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: raphlinus</title><link>https://news.ycombinator.com/user?id=raphlinus</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 18 Nov 2026 04:49:54 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=raphlinus" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by raphlinus in "Large Angle Period of the Pendulum"]]></title><description><![CDATA[
<p>What, no elliptic integrals?<p>It turns out the math for the pendulum is the same as the math for the elastica. I go into this in some depth in my history of the elastica [1].<p>[1]: <a href="https://www2.eecs.berkeley.edu/Pubs/TechRpts/2008/Archive/EECS-2008-103.pdf" rel="nofollow">https://www2.eecs.berkeley.edu/Pubs/TechRpts/2008/Archive/EE...</a></p>
]]></description><pubDate>Tue, 17 Nov 2026 03:14:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=50526925</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50526925</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50526925</guid></item><item><title><![CDATA[New comment by raphlinus in "Squircles"]]></title><description><![CDATA[
<p>I think a rounded rectangle is defined as 4 straight line segments joined with G1 continuity to quarter circle arcs. I'm unaware of any other definition. The only variation I know is whether the radii have to be the same or can be different. Flexible implementations do allow the latter.</p>
]]></description><pubDate>Mon, 16 Nov 2026 21:53:19 +0000</pubDate><link>https://news.ycombinator.com/item?id=50523868</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50523868</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50523868</guid></item><item><title><![CDATA[Squircles]]></title><description><![CDATA[
<p>Article URL: <a href="https://linebender.org/wiki/curves/squircle/">https://linebender.org/wiki/curves/squircle/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50515151">https://news.ycombinator.com/item?id=50515151</a></p>
<p>Points: 223</p>
<p># Comments: 56</p>
]]></description><pubDate>Mon, 16 Nov 2026 07:55:10 +0000</pubDate><link>https://linebender.org/wiki/curves/squircle/</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50515151</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50515151</guid></item><item><title><![CDATA[New comment by raphlinus in "FM-1 Workbench"]]></title><description><![CDATA[
<p>Yes, it's all software, and FM synthesis is using the same engine as dexed. It's a Blackfin chip, which has pretty good throughput for DSP (digital signal processing). A ton more detail is at [1], including the operating system, which is a FreeRTOS-derived SMP kernel.<p>[1]: <a href="https://github.com/AL-255/FM-1-RE" rel="nofollow">https://github.com/AL-255/FM-1-RE</a></p>
]]></description><pubDate>Fri, 13 Nov 2026 21:29:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=50490037</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50490037</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50490037</guid></item><item><title><![CDATA[New comment by raphlinus in "Rust 1.100"]]></title><description><![CDATA[
<p>I get your point but disagree strongly. One topic that's been on my mind a lot is durability. The culture of LLMs is that everything is transitory. Why bother keeping a vibe coded codebase around for more than a few weeks, when the next big model iteration will be able to vibe code it better?<p>By contrast, free software in general, and Rust very much in particular, has cultural norms that things should last a while, rather than immediately being thrown away. I think that's worth preserving.</p>
]]></description><pubDate>Thu, 12 Nov 2026 20:15:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=50473513</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50473513</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50473513</guid></item><item><title><![CDATA[New comment by raphlinus in "A comprehensive study of how fingerprintable users are"]]></title><description><![CDATA[
<p>If you can run compute shader code, the fingerprinting and side channel possibilities become very rich. [1] is one paper among many on this topic.<p>[1]: <a href="https://dl.acm.org/doi/10.1145/3634737.3656283" rel="nofollow">https://dl.acm.org/doi/10.1145/3634737.3656283</a></p>
]]></description><pubDate>Tue, 10 Nov 2026 21:59:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=50446881</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50446881</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50446881</guid></item><item><title><![CDATA[New comment by raphlinus in "CUI – Native UI components for C"]]></title><description><![CDATA[
<p>For this experiment I had it write directly.<p>I should add, part of the experiment was to test whether having variable names and (C style) types would help the LLM. For a task of this size (about 1.5 kLoC Rust, 3k insn asm) it was inconclusive, but for larger projects I suspect it would make a difference.<p>And the plan for symbolasm is to move past C style types into Rust style and maybe dependent typing, specifically to facilitate automated proofs.</p>
]]></description><pubDate>Mon, 09 Nov 2026 21:03:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=50432088</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50432088</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50432088</guid></item><item><title><![CDATA[New comment by raphlinus in "CUI – Native UI components for C"]]></title><description><![CDATA[
<p>Yes, and? As the cost of code generation approaches zero, this starts to make more sense, especially if you <i>do</i> have mechanisms in place to ensure safety and correctness.<p>I did an experiment a couple weeks ago of having Opus 5.5 translate an MP3 decoder from Rust into Cortex-M33 assembly language, both raw asm and symbolasm. The code was 3x smaller and ~1.5x faster than Rust -O3. These seem like gains that are worth it.</p>
]]></description><pubDate>Mon, 09 Nov 2026 20:41:13 +0000</pubDate><link>https://news.ycombinator.com/item?id=50431761</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50431761</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50431761</guid></item><item><title><![CDATA[New comment by raphlinus in "GeoWorks, Microsoft Windows' Upstart '90s Competitor (2016)"]]></title><description><![CDATA[
<p>That's quite cool. I haven't dug deep into using the PIO but it clearly has potential, as shown by Linus Åkesson's demo work.</p>
]]></description><pubDate>Sun, 08 Nov 2026 22:47:13 +0000</pubDate><link>https://news.ycombinator.com/item?id=50418314</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50418314</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50418314</guid></item><item><title><![CDATA[New comment by raphlinus in "JIT bitmap text rendering for microcontrollers"]]></title><description><![CDATA[
<p>Thanks! I had a feeling there was something out there, but couldn't easily find it. I can see this is well optimized for speed, but the RAM usage looks scary from the perspective of a really memory constrained chip.<p>On the RP2350, a load from memory and a move-immediate both take 1 cycle, because there's a separate data bus to fast SRAM. If I did want more speed and didn't mind an expansion in font size, I could store each glyph in its own byte/word and avoid the ubfx instruction. But on this chip, for a reasonable speed/space tradeoff I'm going to optimize for space every time.</p>
]]></description><pubDate>Sun, 08 Nov 2026 20:27:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=50416812</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50416812</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50416812</guid></item><item><title><![CDATA[New comment by raphlinus in "GeoWorks, Microsoft Windows' Upstart '90s Competitor (2016)"]]></title><description><![CDATA[
<p>Hi! I'm doing something similar: vic-2350. But I'm not porting any existing system, I'm doing it to match the constraints of the hardware we have now.</p>
]]></description><pubDate>Sun, 08 Nov 2026 02:32:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=50409964</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50409964</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50409964</guid></item><item><title><![CDATA[JIT bitmap text rendering for microcontrollers]]></title><description><![CDATA[
<p>Article URL: <a href="https://linebender.org/blog/jit-bitmap-text/">https://linebender.org/blog/jit-bitmap-text/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50408729">https://news.ycombinator.com/item?id=50408729</a></p>
<p>Points: 5</p>
<p># Comments: 1</p>
]]></description><pubDate>Sat, 07 Nov 2026 22:41:52 +0000</pubDate><link>https://linebender.org/blog/jit-bitmap-text/</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50408729</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50408729</guid></item><item><title><![CDATA[New comment by raphlinus in "The Flash C Compiler"]]></title><description><![CDATA[
<p>I think you're right, but rather surprisingly I have some experimental data on this question that is inconclusive at best. My hobby project is symbolasm, which is intended to close exactly the gaps you describe: by naming variables and introducing types, it re-adds much of the context that assembly language erases. I ran a quick'n'dirty experiment to see if it would make the LLM more efficient at writing code than raw assembler.<p>The task was rewriting an MP3 decoder from Rust into ARM assembly language, Cortex-M33 flavor specifically, optimizing for both speed and code size. The output is approx 3k instructions. The tl;dr is that Opus 5.5 nailed the task, whether writing assembly or symbolasm. It took approximately 2x tokens for the latter, which I think is largely explained by the fact that it was dealing with a novel language nowhere represented in its training set.<p>The quality of the result was comparable; the raw asm was slightly faster and slightly less dense, which I consider insignificant because I think you can get the same effect just by tuning the prompt. In particular, if I were re-running the experiment, I might set a strict code size budget and then ask it to optimize for speed within that budget, allowing comparison on a single scalar.<p>The experiment was compelling in another way: the code was 3x denser and 1.5x faster than -O3 output from Rust. So for certain applications, going to assembler <i>will be</i> worth it, especially as the cost of generating the code goes to zero.<p>[1]: <a href="https://github.com/raphlinus/symbolasm" rel="nofollow">https://github.com/raphlinus/symbolasm</a></p>
]]></description><pubDate>Sat, 07 Nov 2026 20:35:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=50407806</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50407806</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50407806</guid></item><item><title><![CDATA[New comment by raphlinus in "AI is forcing us to pay off 40 years of technical debt"]]></title><description><![CDATA[
<p>No, it's Knuth. The exact quote is "Beware of bugs in the above code; I have only proved it correct, not tried it." [1]<p>[1]: <a href="https://staff.fnwi.uva.nl/p.vanemdeboas/knuthnote.pdf" rel="nofollow">https://staff.fnwi.uva.nl/p.vanemdeboas/knuthnote.pdf</a></p>
]]></description><pubDate>Thu, 05 Nov 2026 08:32:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=50376659</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50376659</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50376659</guid></item><item><title><![CDATA[New comment by raphlinus in "Partial Operator Precedence"]]></title><description><![CDATA[
<p>Worth noting, WGSL does this too, and is mostly a bit stricter than Zig: `a + b & c` and `a << b & c` are errors in WGSL but accepted by Zig. (Quick: off the top of your head, how do those parse)?<p>I'm thinking of doing this for symbolasm also.</p>
]]></description><pubDate>Thu, 05 Nov 2026 08:08:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=50376481</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50376481</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50376481</guid></item><item><title><![CDATA[New comment by raphlinus in "Polyend releases the Tracker firmware source code"]]></title><description><![CDATA[
<p>They should send me a t-shirt. But in any case, thanks!</p>
]]></description><pubDate>Tue, 03 Nov 2026 10:31:28 +0000</pubDate><link>https://news.ycombinator.com/item?id=50350235</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50350235</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50350235</guid></item><item><title><![CDATA[New comment by raphlinus in "Nibnib: Image compression with extreme decompression speed"]]></title><description><![CDATA[
<p>Both the image compression and the symbolasm language started out being entirely hand coded, but I have used Claude to move it forward. I have complicated feelings about this. Basically I would have gotten these results even without the LLM, but it would have taken a lot longer.<p>I feel like one of the main tasks in front of us as an industry right now is to figure out how to use LLMs <i>well,</i> how to create quality and resist slop. As part of that, I'm taking care to write the prose myself.<p>I totally understand how any LLM usage can be a turnoff, though. I've felt that myself.</p>
]]></description><pubDate>Mon, 02 Nov 2026 05:59:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=50333572</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50333572</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50333572</guid></item><item><title><![CDATA[New comment by raphlinus in "Nibnib: Image compression with extreme decompression speed"]]></title><description><![CDATA[
<p>I don't believe macro assemblers do a flow-based analysis to verify that the register allocation for variables is consistent; from what I understand the most you get is just a scoped alias for the name of the register. Of course, I could be missing something. If you know of anything beyond the relevant work I cite, please let me know.</p>
]]></description><pubDate>Mon, 02 Nov 2026 02:56:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=50332618</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50332618</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50332618</guid></item><item><title><![CDATA[New comment by raphlinus in "Nibnib: Image compression with extreme decompression speed"]]></title><description><![CDATA[
<p>If you have an Adafruit Fruit Jam, you can run the demo on that board [1]. It can stream 640x480 video off an SD card at 60fps, as well as display a static image from flash.<p>Also possibly of interest to HN, the decoder[2] is written in a new low level language called symbolasm[3], which seeks to combine the advantages of assembly language (it's possible to reason about and optimize performance with few limits) with some of the affordances of C, especially named variables and struct fields.<p>[1]: <a href="https://github.com/raphlinus/vic-2350" rel="nofollow">https://github.com/raphlinus/vic-2350</a><p>[2]: <a href="https://github.com/raphlinus/vic-2350/blob/main/image.sy" rel="nofollow">https://github.com/raphlinus/vic-2350/blob/main/image.sy</a><p>[3]: <a href="https://github.com/raphlinus/symbolasm" rel="nofollow">https://github.com/raphlinus/symbolasm</a></p>
]]></description><pubDate>Mon, 02 Nov 2026 01:10:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=50331792</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50331792</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50331792</guid></item><item><title><![CDATA[New comment by raphlinus in "Nibnib: Image compression with extreme decompression speed"]]></title><description><![CDATA[
<p>Absolutely. This is optimized for highly resource constrained devices, not modern machines. I can tweak the README if that's at all confusing.</p>
]]></description><pubDate>Sun, 01 Nov 2026 21:35:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=50330143</link><dc:creator>raphlinus</dc:creator><comments>https://news.ycombinator.com/item?id=50330143</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50330143</guid></item></channel></rss>