Wow that entire workflow looks absolutely hellish

> for why mathematicians should widely receive funding for merely understanding things

This is all they were receiving funding for previously? Nothing has particularly changed there.

Incredible, a post and repo dedicated to a markdown file, the downfall of this field has been swift.

Did you open the repo? There's a subdirectory with two dozens of files, around 300Kb of text.

Storing/visualizing small text changes over time as revisions is exactly what Git is excellent at, how else would you keep track of updates to the prompts?

> how else would you keep track of updates to the prompts

I dont? because prompts are not a thing that are ever needed to be tracked lol.

but they're very huge markdown files. look how much junk they're polluting the conext window with.

yes they are incredibly unpleasant to work with. Like the most annoying coworker you ever had.

For my own sanity ive wrapped my pi setup in some strict bubble wrap and it is basically exclusively an LOC generator. I have no need nor desire to "converse" with the LLM, it is there to do exactly what i need when i need it.

> Or maybe test coverage has/will become too noisy a parameter to use at all.

It already is, ive banned unit tests via ci checks from our codebases, they were not particularly useful before LLMs and now they are a net negative.

We require int and some e2es and that does all that units do and more.

I sometimes use unit tests to fill in gaps. I just caught a function with almost full coverage except for an if statement.

Playwright e2e tests are harder to maintain than pytest code, so I wrote a test just for that one case, and while I was at it, a few other input strings that seemed like they could be problems.

If the function had no test coverage at all, I probably would have tried to get it with an integration test, because that would mean the callers also weren't getting tested.

The occasional unit test is fine. But you do not need for example `test_helper_test.go` that is just ridiculous.

Its gotten so bad I decided to throw the baby out with the bathwater so to speak and focus on the high value places instead of trying adjudicate every individual usage

Are they slow and brittle?

I like having both. Unit tests are a decent indicator for local development. Fast, quick to iterate on. Integration tests are slow as molasses and I can’t iterate with them when it takes 20 minutes to setup the suite and run everything. Too slow. But still useful as you say.

They don't flake, generally can run a few hundred in 3-4 mins. Iteration times are definitely key.

Im not TYPING a lot of code by hand anymore (still some, inference is slower than typing in a few cases) but im still in the shell zipping around doing other stuff like i always have, the LLM is helpful in a lot of ways but it in no way shape or form is the ideal way to interact with a computer.

So i still get daily use out of these tricks.

Are there people that are literally ONLY interacting with a computer via an LLM? thats crazy if its true

I dont think an article title "dropping ebpf cpu cost with memoization" really needs to be accessible to a broad audience.

Thx claude