sceadu

this has been the exact analogy I've kept in my mind with LLMs

not to tell you that it's better or worse but you might like helium if you are using ungoogled chromium

deal for he, heel for thee

by concentrating the load into a single location like that, you would likely overload the transmission system at that location and trigger a ton of system upgrades. so they would likely need to build their own generation and still probably upgrade lines if they just took over a county.

I really hate SQL as a language for doing data-related tasks, but the core tech for it is often much better in terms of reliability (and is close to having performance parity IME)... specifically for working with larger-than-memory datasets. I think the team at Polars is working on improving larger-than-memory operations though.

In my experience it works OK with spilling to disk so I haven't had too much of a concern with memory usage... previously I had issues with it OOM'ing and failing (or maybe this was a skill issue?), but haven't had that happen recently.

the I in ACID stands for isolation, not idempotency

dang I had no idea you wrote the asof join for pandas. thank you for that

I usually do REPL driven development in Python via emacs but you can tell it's not the natural way to do things, esp. if you start doing stuff with async. But I still feel that it makes me way more productive than I would otherwise be

Always Increasing

You might be able to do it with ibis. Don't know about the performance though