Fixpoint

2026-08-03

#jwrd logs for Aug 2026

Filed under: #jwrd logs, Logs — Jacob Welsh @ 20:34
Day changed to 2026-08-03
[20:34] nekoluce: guys coldcards are being attack according to news
[20:34] nekoluce: is not the same as btc wallets right?
[22:09] dorion: coldcard is/was a vendor of btc wallets.
[22:24] nekoluce: https://newsroompanama.com/2026/08/03/the-first-water-fountain-generated-by-atmospheric-humidity-was-unveiled-at-omar-park-in-panama/
[22:50] nekoluce: are you guys okay because of today attack on the cold cards?
[23:16] jfw: they were supposedly a competitor, for people who didn't want to do their JWRD homework.
[23:17] jfw: CoinKite was the company, if I recall.
[23:27] dorion: http://jfxpt.com/2026/jwrd-logs-for-aug-2026/#15858 -- I use gbw with the fg trng, no sweat here.
[23:27] wotbot: 2026-08-03 22:50 (#jwrd) nekoluce: are you guys okay because of today attack on the cold cards?
Day changed to 2026-08-07
[02:10] nekoluce: https://cybernews.com/news/russian-hackers-hotel-wifi-malware-midnight-blizzard/
Day changed to 2026-08-08
[14:54] nekoluce: https://www.youtube.com/watch?v=qMoj-njeuLA
[20:29] whaack: sry dorion for not responding to you on heathen comms, ive been traveling for the last fwe months, i may have time to fire up eulora soon
Day changed to 2026-08-10
[17:38] nekoluce: eulora rules!
Day changed to 2026-08-11
[20:55] nekoluce: https://cryptoticker.io/en/russia-crypto-law-us-clarity-act-delay/
[22:01] nekoluce: https://acp.copernicus.org/articles/25/2937/2025/?utm_source=substack&utm_medium=email I suggest air purifiers at home because of the desert of sahara dust and also this new polution in the strat of malaca
Day changed to 2026-08-18
[00:02] nekoluce: https://www.youtube.com/watch?v=X_prH2KAwVQ sam altman speech
[02:56] jfw: nekoluce, you keep messaging me then quitting.
[03:08] jfw: and otherwise, why should we listen to sam altman speeches?
Day changed to 2026-08-23
[00:32] nekoluce: https://www.youtube.com/watch?v=Kw4zV3_vPko 11:39 bunkers
[00:36] nekoluce: https://www.youtube.com/watch?v=sEBPFBB-5-M&list=RDsEBPFBB-5-M&start_radio=1 neoclassical electronic music :)
Day changed to 2026-08-24
[16:59] nekoluce: https://www.youtube.com/watch?v=IAIqJfNTdMA what do you guys think about this countries ?
[19:30] nekoluce: https://www.youtube.com/watch?v=sF60iCzbjv4 philosophy
Day changed to 2026-08-26
[16:43] jfw: Gales Linux 26.1 patch release is out and signed in the usual place, with the promised fixes and more
[16:43] wotbot: 2026-03-19 20:10 (#jwrd) jfw: http://jfxpt.com/2026/jwrd-logs-for-Mar-2026/#15590 - it's already going to need a patch release, as a syntax error in /etc/rc snuck past the eagle type eyes of the QA department. Apparently they only tested system updates, not the full install from zero. It doesn't break the boot but does defeat the new feature, which was meant to set up a /dev/rtc symlink to work ar
[16:43] wotbot: 2026-04-12 05:11 (#jwrd) jfw: also in pending Gales updates is a needed PHP fix for untangling the executed script path from trailing URL path info; for now, the only secure configuration for fastcgi interfacing is to set cgi.fix_pathinfo=0 *and* cgi.discard_path=1 in /etc/php.ini.
[16:50] jfw: including a new hardened port of expat the xml parser toolkit, which replaces the old vulnerable version bundled in python and enables the xml extension in php; further improvements to clockspeed, djbdns, gales-util, httpd, pcre, php 5.6 and python 2.7 ports; a fix for tmux breaking the console line drawing/graphics characters; and some minor kernel drivers enabled for useful hardware.
[17:00] jfw: one that didn't make it on this round is that I discovered python's .pyc compilation cache invalidation mechanism was nondeterministic and broken with respect to package distribution: it records source file modification timestamps in .pyc headers, as the cheap substitute for detecting file content changes. thus it improperly rejects packaged .pyc files because such build-time metadata is
[17:00] jfw: deliberately excluded as meaningless in Gales packaging; this results in unnecessarily slow startup or import of system modules as an unprivileged user if they haven't previously been imported by root.
[17:03] jfw: I'm planning to fix this by replacing that mtime with full source md5 hash and possibly file size in the .pyc header.
[17:16] jfw: md5 is the right choice here because cryptographic strength is not needed - a collision attack would only apply at the point that you've already decided to execute the attacker's code; yet something with a stronger avalanche design than crc32 is still in order, to rule out accidental collisions from one change counterbalancing another; and of the well known hash functions that I've implemented,
[17:16] jfw: md5 is the fastest.
[17:25] jfw: python 3 already adopted some strange leaky hybrid variant of this idea, https://peps.python.org/pep-0552/
[17:29] jfw: I don't get it, they're happy to break the entire ecosystem of source code by renaming unicode as string and string as bytes, but they're afraid to invalidate fragile cache files which will regenerate anyway.

No Comments »

No comments yet.

RSS feed for comments on this post. TrackBack URL

Leave a comment

Powered by MP-WP. Copyright Jacob Welsh.