lvncelot

One addition: Postgres for all your GPU-based machine learning training and inference needs: https://github.com/postgresml/postgresml

> If the conversation feels risky or off, saying less and giving shorter replies is safer and less likely to cause harm.

Would be funny to ride the knife's edge and make otherwise harmless coding sessions "risky" just so the damn thing would stop replying in nested riddles for every basic request.

Imagine what could have been...

Ohh fnox looks really cool, with encryption being one possible provider but something like Vault being another. Thanks for the recommendation.

Interesting to see that PyTorch has become the de-facto standard after it seemed for a while that the dust had settled on Tensorflow.

I also thought that Jax would in turn take over after PyTorch but it never seemed to quite take off (still in use though from what I can tell).

I'm honestly more surprised that they could resist the temptation to call it Copilot

> It's the same class of bug as manually parsing HTML with regex, it works right up until it doesn't

I'm sure you already know this one, but for anyone else reading this I can share my favourite StackOverflow answer of all time: https://stackoverflow.com/a/1732454

I blocked them from my suggestions page with ublock origin

Yeah, there's a good reason why I have a lot of criticisms about CSS and only some about, say, QT. It's because I spent the last 15+ years writing CSS in some way or another and one extended summer of hacking a side project in QT, abandoning it, and never looking back.

Hm. Just a few months back, the KiCad team posted about their woes with Wayland: https://www.kicad.org/blog/2025/06/KiCad-and-Wayland-Support...

Adding https://github.com/firefly-iii/firefly-iii

Completely agree. In uni, I (re)-learned about vectors in linear algebra, and for a good chunk of the course, we didn't write anything in "standard vector notation". We learned about vector axioms first, and then vectors were treated as "anything that satisfies the vector axioms". (When doing more practical examples, we just used the reals instead of something like R^3, but the entire time it was clear that for any proof, anything that can be added and multiplied in the way that the vector axioms describe would fit.) I think adopting this structuralist view really helps with a lot of mathematical studies.

> cut off ChatGPT in Germany

God I can only hope

You can achieve both a clean history and granular state in the URL with using history.replace() and history.push() where necessary.

Honourable mention: https://jspaint.app

Aren't they're used for slightly different things though? GIMP for image manipulation and Krita for digital drawing?

(Krita is pretty awesome though, it's up there with Blender for me)

That last one has to be a troll, holy shit.

I think more often than not, companies are using a single cloud provider, and even when multiple are used, it's either different projects with different legacy decisions or a conscious migration.

True multi-tenancy is not only very rare, it's an absolute pain to manage as soon as people start using any vendor-specific functionality.

Same boat, and if anything AI has made the "disentangling spaghetti code" part even more annoying. In the past, badly designed code at least hat the decency to also look bad on first glance.

Every time I read an apple press release, I immediately bounce off because of the purposeful omission of the definite article when referring to their products; like "iPhone Air features ..." instead of "The iPhone Air features ...".

It's irrational, but it's like an uncanny valley via text for me.

Good looking phone though.

Ha, I actually thought about getting a Boox when I started spending some time in my garden to work this summer cause of that exact same reason. Good to hear that it's actually feasible.

Two things:

1. The user hostile changes I mentioned were a degradation of the initial experience, i.e. herding people into Remarkable's Cloud offering which, while it allowed for grandfathering, still was very restrictive in what sync features are available. Also, for new customers buying a RM now means a monthly subscription cost, which is why I said can't recommend it.

2. Remarkable itself was using the openness of their tablet in their marketing. If I were to buy an iPhone and then complain about the walled garden, that would be one thing. If I buy a product that prides itself as being hackable, I don't think I'm wrong to expect that.

Lastly, saying "I can't recommend this because of XYZ" is a far cry from "trashing the company".

I have a Remarkable 2 that I used to use religiously, now use sporadically, but cannot recommend because of the user-hostile changes to the subscription and the very restrictive underlying software.

One of the promises that lead me to buy one was the hackability - "It's Linux!" "You can SSH into it!", which, on paper (heh) is still true, but in practice very much isn't.

I think something like a Boox, which runs Android, might be more open to customization, but for now I am back to pencil and paper. That doesn't run Linux, but it also won't change its terms of service anytime soon.

After hearing them myself, I think I know what you mean. The voices get a bit warbly and sound at times like they are very mp3-compressed.

I think current theories break down at less than a Planck length, but they are not constrained to integer multiples of it.

It's funny how, for a certain subset of math, a researcher's life can be condensed to "arguing about which number is the biggest (preschool)" -> "learning about math" -> "arguing about which number is the biggest (academia)"

"Gotcha, here's an XML response"

I'll have to take another look but I always thought that the Python type experience was a bit more clunky than what TS achieved for JS. I guess there's also a critical mass of typing in packages involved.

I was talking to a friend recently about how there seem to be less Vue positions available (relatively) than a few years ago. He speculated that there's a feedback loop of LLMs preferring React and startups using LLM code.

Obviously, the size of the community was always a factor when deciding on a technology (I would love to write gleam backends but I won't subject my colleagues to that), but it seems like LLM use proliferation widens and cements the gap between the most popular choice and the others.

0. Race conditions