robertfw

This was happening even on the `stable` branch with 4.7

I managed to get claude to create a recovery script to un-brick sessions, YMMV

https://gist.github.com/robertfw/993dbe8643c4fbdf12005dff2ec...

OpenAI was doing this back with GPT2, saying it was too dangerous to release

I was in this camp as well until recently, in the last 2-3 weeks I've been seeing problems that I wasn't seeing before, largely in line with the issues highlighted in the ticket (ownership dodging, hacky fixes, not finishing a task).

how utterly disconnected from reality you are

`uv` has per-package overrides, I imagine there may be similar in other managers

Yeah, the moment I saw this I knew this article was not going to be very useful.

Getting claude to follow your guidance files consistently is a bit maddening.

I really don't get this sentiment. The only sets that I think didn't contribute like this were the bionicle stuff. Getting a few more unique parts with a set gives you more options, not less.

I wonder if it's just: compact earlier, so there's less to compact, and more remaining context that can be used to create a more effective continuation

I've been using a `/feedback ...` command with claude code where I give it either positive or negative feedback about some action it just did, and it'll look through the session to make some educated guesses about why it did some thing - notably, checking for "there was guidance for this, but I didn't follow it", or "there was no guidance for this".

the outcome is usually a new or tweaked skill file.

it doesn't always fix the problem, but it's definitely been making some great improvements.

https://www.lego.com/en-ca/product/lego-large-creative-brick...

I was already programming when I got into mindstorms (so very happy that my high school had an "Industrial Control Technologies" class), but the jump in complexity and the thrill of seeing my code power something physical was definitely a turning point

I still have my mindstorms controller + sensors, I wonder how much effort it would be to get everything up and running again

I feel like the special parts have eased off, it was pretty bad with the bionicle stuff (which ironically are apparently what saved lego from financial difficulties), but I'd say all of the recent sets I've got (I get one a year for Christmas) going back at least 5 years have been made up of relatively generic parts, with the odd little special bit for flair.

I do think that the "model" form of something recognizable (thinking mostly about the Star Wars kits here) does lend itself to kits that sit on the shelf, but no one is forcing you to keep it assembled.

With that said, one thing I remember seeing in older kits is instructions for more than one build, which could serve as a kicking off point for someone to reapproach a bundle of bricks in multiple ways.

My own guideline is that once something is built, after a little while it has to be either modified/evolved, or disassembled and put into the bucket-o-bricks for reuse (well if I'm honest, it's several very well sorted craft/hardware bins)

there are going to be quite a lot of places where getting a laptop is a considerable expense

and if your pre-commit runs fast (thank you, ruff), you can set this up as a post-edit hook in claude code

I feel like this talk by John Hughes showed that there is real value in this approach with production systems of varying levels of complexity, with two different examples of using the approach to find very low level bugs that you'd never think to test for in traditional approaches.

https://www.youtube.com/watch?v=zi0rHwfiX1Q

You could already do these things before, you just spent much more time twiddling your thumbs waiting for lock files to be resolved

Slow doesn't really begin to do justice, I'd have to wait for >5 minutes for pipenv to finish figuring out our lock file. uv does it in less than a second.

I don't know, I tend to agree. I feel like the number of times I've been thrown off by an out of date comment for code that could have probably been refactored to be clearer, outweigh the times a comment has helped.

Docstring comments are even worse, because it's so easy for someone to update the function and not the docstring, and it's very easy to miss in PR review

comments for me are a code smell:

- like all documentation, they are prone to code rot (going out of date)

- ideally code should be obvious; if you need a comment to explain it, perhaps it's not as simple as it could be, or perhaps we're doing something hacky that we shouldn't

KDE won me over for the simple fact that it's highly configurable, and that configuration is all driven out of one UI tool. Gnome drove me nuts with molding it into the shape I wanted.

my development workflow with python often boils down to repeatedly running `pytest --ff --pdb -x`...

- run my tests, running any previous failures first - drop into the debugger when a test fails (or of course, hits a breakpoint) - end the test run after the first failure

use debugger to figure out what went wrong, fix it, wash, rinse, repeat

I've had similar cases where the fix to the test was.. delete the test. Ah, if only I'd realized that little hack earlier in my career!

Making it work can be a hacky, tech debt laden implementation. Making it right involves refactoring/rewriting with an eye towards maintainability, testability, etc etc

Might as well put that effort towards halting the ongoing mass extinction of more animals, before worrying about bringing back facsimiles of the ones we've already eliminated

it's exceedingly common to use a monochromatic sensor with different filters and then combine them in post. you might use visual filters - e.g., take 3 exposures using red/green/blue filters, but it's also common to use filters for specific wavelengths e.g. hydrogen, oxygen, and sulfur, and then assign RGB to each filter arbitrarily

there's lots of other red flags with this product but I wouldn't care about this one

https://en.wikipedia.org/wiki/Datalog

The jobs seem to be pretty available in Europe, but I've been struggling to find Clojure work over here on the west coast of North America. I've had to bite the bullet and am looking for work in other languages now, though I'd really rather stay in the ecosystem.

I did a stint working on trust & safety tools for a social media startup. What I learned from working with content moderators is that overall, the industry gets by doing the barest of bare minimums, and TikTok is one of the worst offenders

"But the cost!", well, if the business model can't be viable when done safely, then the business model is not viable - we just don't have the externalized costs of harm done visible to help make that call