Hacker News: ondra

New comment by ondra in "Intel Arc Pro B70 Review"

ondra — Wed, 29 Apr 2026 06:38:55 +0000

This is incorrect, prompt processing is compute bound.

New comment by ondra in "Not buying another Kindle"

ondra — Mon, 20 Apr 2026 21:21:39 +0000

Note that Boox does not release the sources for the GPL software they redistribute.

New comment by ondra in "Windows 11 Notepad to support Markdown"

ondra — Wed, 25 Feb 2026 21:08:11 +0000

This might be the reason for Markdown support, LLMs love it.

New comment by ondra in "Linux From Scratch ends SysVinit support"

ondra — Tue, 03 Feb 2026 14:34:06 +0000

> Developers don't have to package their project for all distros

Essentially nobody uses the sources we provide. Literally nobody packages them. A few people use our rpm and deb packages, but the vast majority uses a (slightly broken and outdated) docker image built by third party.

You might not like it, and I certainly do not, but unfortunately containers seem to be the best alternative that just works, compared to everything else.

New comment by ondra in "Rubio stages font coup: Times New Roman ousts Calibri"

ondra — Thu, 11 Dec 2025 16:16:27 +0000

> It was Blinken that arbitrarily introduced

The _second paragraph_ of TFA gives a reason for the introduction. Please explain how you came to the conclusion that the change was arbitrary.

New comment by ondra in "APT Rust requirement raises questions"

ondra — Tue, 25 Nov 2025 16:51:33 +0000

Pentium Pro is the first i686 CPU, so you should be fine.

New comment by ondra in "Evaluating the Infinity Cache in AMD Strix Halo"

ondra — Wed, 22 Oct 2025 12:51:54 +0000

GPT-OSS 120B uses native 4 bit representation, so it fits fine.

New comment by ondra in "Garmin beats Apple to market with satellite-connected smartwatch"

ondra — Sun, 07 Sep 2025 19:46:25 +0000

Works with a Samsung watch and their app. (I wouldn't really recommend it, though.)

New comment by ondra in "Gemma 3 270M: Compact model for hyper-efficient AI"

ondra — Thu, 14 Aug 2025 21:04:16 +0000

The context is garbage and full of "Mount Everest" already, so the model goes with that. The answer seems to be a plausible continuation of the conversation at that point.

New comment by ondra in "Qwen3-235B-A22B-Thinking-2507"

ondra — Fri, 25 Jul 2025 19:35:22 +0000

Yes. The original DeepSeek-R1 answered those questions just fine. The newer models seem to be much more brainwashed.

New comment by ondra in "Show HN: KVSplit – Run 2-3x longer contexts on Apple Silicon"

ondra — Fri, 16 May 2025 20:58:17 +0000

Is this any different from using --cache-type-k and --cache-type-v?

New comment by ondra in "The C23 edition of Modern C"

ondra — Wed, 16 Oct 2024 13:06:43 +0000

MIPS is still quite alive in consumer networking hardware.

New comment by ondra in "Elon Musk reportedly had his own tweets prioritized out of jealousy"

ondra — Wed, 15 Feb 2023 23:47:56 +0000

I block every ad I see, seems to make the rest of them go away.

New comment by ondra in "Gas Leak in the Baltic Sea"

ondra — Tue, 27 Sep 2022 23:35:48 +0000

Biden most certainly did not announce he would do _this_.

New comment by ondra in "Show HN: "q", a DNS query tool with support for UDP, TCP, DoT, DoH, DoQ and ODoH"

ondra — Sun, 17 Apr 2022 23:23:04 +0000

Could be useful, but not _that_ useful for me to consider polluting my global shell namespace with this name.

New comment by ondra in "For what are the Windows A:\ and B:\ drives used?"

ondra — Sat, 15 Jan 2011 08:00:00 +0000

32-bit Windows 7 runs 16-bit programs just fine.

New comment by ondra in "Why GIMP is Inadequate"

ondra — Wed, 12 Jan 2011 04:43:08 +0000

CRTs will show what the RAMDAC on the video card feeds them. They shouldn't have limitations like this.

New comment by ondra in "Google Chrome for Linux goes stable"

ondra — Wed, 26 May 2010 04:54:18 +0000

I'm a fan of Konqueror myself. For some reason nobody seems to use it. Konqueror is stable, fast, doesn't use much memory and renders everything just fine.