2026-07-01 02:47:13 abandoning git because it uses static linking with Rust now smacks of the deepest paranoia 2026-07-01 02:49:40 anyways, the only true way to make a machine secure is to remove all cables, put it in a faraday cage, pour concrete around the faraday cage, and drop it in the marianas trench 2026-07-01 02:49:56 or blow it to smithereens 2026-07-01 02:50:08 then your code and data will be secure 2026-07-01 06:17:41 First, "security" depends on your threat model. For some threat models, airgapping the machine makes a lot of sense; for others, not so much. Consider that, for instance, that computer viruses thrived in largely Internet-less world, then became extinct once Internet became commonplace. 2026-07-01 06:17:41 Or consider this scenario: people buy cheap "Wi-Fi routers" for their homes. These routers are often enough easy to crack into. The attackers exploiting the vulnerabilities are not interested in users' /data/ - but rather in their IPv4 addresses. Once they put their code on enough routers, they can DDoS, send junk mail, wreak havoc on IRC channels, etc. 2026-07-01 07:44:16 There is something to be said about the million dependencies that get pulled into Rust just for basic language features and libraries that have zero dependencies in C. It's cool you can do that, but also dependencies aren't isolated chunks of functionality, they're porous and security is one of the many holes 2026-07-01 07:45:10 And Rust sort of acts like you should just use whatever you want, makes it really easy to pull it all in, but the reality is that having lots of dependencies is *hard* and they've not solved the bigger problems there, just the easy ones (of managing the build, configures, etc) 2026-07-01 07:45:40 And we see e.g. with JavaScript, which is also memory-safe, the security issues they've had from similar centralised package systems 2026-07-01 07:45:52 And similar fragmented libraries 2026-07-01 07:46:45 It's something that makes Forth interesting in 2026 because the opposite culture exists in Forth, a lot of people will "do it yourself" with things that you'd usually use a battle-worn fundamental C library for 2026-07-01 08:07:31 So far as I can tell, such cultures aren't exclusive to Forth and Rust, respectively, but the difference tends to be the more pronounced the greater the gap in the platforms' adoption is. Other than that, other things being equal, the more the code - the more the bugs. "Classic" Forth dialects tend to disadvantage code getting too big, which, indirectly, helps keeping it more secure. 2026-07-01 08:12:44 And there's also the "human factor" to consider: different people are prone to introducing different bugs into the code. Ideally, a programmer should choose a platform that helps them avoid the bugs that cause the most trouble for /them./ Which do not necessarily have anything to do with memory safety. 2026-07-01 08:15:15 ACTION listens to http://he3.magnatune.com/all/12-Street%20Walk-Paolo%20Pavan.mp3 , does not go for a walk yet  2026-07-01 08:35:47 JFTR, I've asked on #pkgsrc, and the regulars there find it likely that pre-Rust version of Git will remain in pkgsrc for a while. (FWIW, pkgsrc still offers GNU Emacs 21 from 2005, though I guess that might be due to the license change than for purely technical reasons.) I suppose that might be enough for me. 2026-07-01 10:20:49 crc: From what I've read, LLM training robots these days cause a lot of traffic to sites that otherwise don't get much. Not something that affected me personally (I run Apache and Lighttpd - and mostly just serve "static" files at that), but others report their servers struggling with the increased load. 2026-07-01 22:49:36 crc i am writing a cellular automata workshop with forth editar for the rules https://ibb.co/842BzdRX 2026-07-01 22:50:43 i'm using the pdf book copy a the margolus and toffoli book about the cam machine