@ThePrimeagen I do that with a simple script that calls `gh pr`, and an Emacs function to pipe the output into a compilation mode buffer. So I can see all the comments from github from both agents and humans and jump to every location very easily, game changer.
@mhmazur I have 3 worktrees and I rotate between them, it's useful specially for things that take a while to run and experiments, I don't want to be blocked while Claude does something that might not go anywhere anyway.
@burkov That assumes that the tests are great, and cover all the behaviours that people rely on. Once something is deployed and has users just rewriting the whole thing to fix a bug might be a very bad idea
And your bloodline dies out, and you never truly grow up. And you have no stake in the future. That’s not a life hack. It would have been unthinkable to boast about this throughout nearly all of human history. Bragging about not having kids because you can travel to Europe more easily and disappear on weekends is about as narcissistic as one can be.
@potetm I use make and bash to be clear when, but in clojure jobs I would use babashka (or python script) since bash is a disgusting programming language for annoying longer than 5 lines of code. It's convenient but let's not pretend that it's a good language
@potetm Make is great, available everywhere, easy to write/read and you also get nice auto completions of the targets with any decent shell. And you can still make it call a script if there is too much logic.
If you have a Python monolith, consider using https://t.co/MnGAhexQCt
and converting your monolith into a monorepo.
uv is brilliant and with workspaces, extracting internal libraries is a breeze.
@potetm It's not about number of chars, recur is a hack that was necessary because you could but do TCO on the JVM I believe. With Haskell recursion is just more natural and the data structure is defined much more clearly