Observations:

4.6 had previously failed to the point where I had to wipe context. It must have written memories because it was referring to the previous conversation.

As the article points out, 4.6 went out of its way to be lazy and came up with an unusable plan. It did extra planning to avoid renaming files (the toplevel task description involves reorganizing directories of files).

4.6 took twice as long to respond as 4.5.

I’m treating this as a model regression. 4.6 is borderline unusable. I’ve hit all the issues the article describes.

Also, there needs to be an obvious way to disable memory or something. The current UX is terrible, since once an error or incorrect refusal propagates, there is no obvious recovery path.

Anyway, with think set to high, I see drastically different behavior: much slower and much worse output from 4.6.

JS is gross, but 16ms (time you get to render a frame at 60 fps) is an eternity on modern systems.

It’s tens of millions of single-threaded CPU cycles.

Also, you probably can use GPU acceleration for client code. That’s enough time for a 2026 integrated CPU to do tens to hundreds of billions of tensor ops.

And yet, the iOS keyboard (presumably multithreaded and native) cannot reliably echo keystrokes in under a second. I regularly see webpages take multiple seconds to redraw a screen.

https://sqlite.org/wal.html

The containers would need to use a path on a shared FS to setup the SHM handle, and, even then, this sounds like the sort of thing you could probably break via arcane misconfiguration.

I agree shm should work in principle though.

The easiest approach is to kill sqlite, then start the new one. I’d use a unix lockfile as a last-resort mechanism (assuming the container environment doesn’t somehow break those).

I challenge you to name one that cannot and that also makes it into high school curricula or How Things Work.

https://mst3k.fandom.com/wiki/A_Case_of_Spring_Fever_(short)

https://m.youtube.com/watch?v=vzKfAFsbRSk

If you are not ready to lock yourself in a bunker after reading the article and watching that short, I strongly suggest you consider the inclined plane.

You’d better do it now. Very few locks work in the absence of transformers, springs and inclined planes.

Most of my commits (hand written and AI) have delete counts that are 75-110% the added line count.

The point that many developers will probably forget to tell the LLM to run cleanup/refactoring paths is probably true though. (I’ve definitely found ghost-chasing bugfixes in all sorts of corners of LLM generated code).

> 'the term “exceeds authorized access” means to access a computer with authorization and to use such access to obtain or alter information in the computer that the accesser is not entitled so to obtain or alter;'

The problem, of course, is that by clicking on a LinkedIn link, you agree to a non-negotiated contract that can change at any time, and that you have never seen. If that weren't allowed, then this sort of crap would correctly be considered "unauthorized access":

https://www.law.cornell.edu/uscode/text/18/1030

I'm happy to see that this doesn't hit firefox. I wonder if safari is impacted.

https://www.eeoc.gov/prohibited-employment-policiespractices

LinkedIn's scanning for browser extensions used by protected groups allows them to provide illegal services to US-based recruiters. I have no idea if they actually do it or not, and am not a lawyer, but common sense suggests there's enough here for a class action suit to move into discovery.

(That'd be amazing if it's possible to do stuff like dump configs + check them into git from the cli, then stand them up on any bhve/sylve box later...)

I've been repeatedly burned by systemd, both on machines I've administered and on appliances. In every situation, the right fix was either "switch distros" or "burn developer-months of work in a fire drill".

In fact, I just decided to go with FreeBSD instead of proxmox specifically because proxmox requires systemd. The last N systemd machines I've had the misfortune to touch were broken due to various systemd related issues. (For large values of N.)

I assume that means anything built on top of it is flaky + not stable enough for production use.