Felipe O. Carvalho (@_felipe) 's Twitter Profile
Felipe O. Carvalho

@_felipe

@ApacheArrow / Databases / Compilers @dbt_labs. (past @SDFLabs, VoDa, @Spotify). C++/OCaml/TLA⁺🇧🇷 → 🇸🇪 → 🌎 bento.me/felipe

ID: 15040668

linkhttp://felipe.rs calendar_today07-06-2008 19:07:44

24,24K Tweet

4,4K Followers

2,2K Following

Felipe O. Carvalho (@_felipe) 's Twitter Profile Photo

`git fixup` is the missing git command I always needed. I will try `git conflicting` next time I see git rebase conflicts. github.com/dinsandra/git-…

Felipe O. Carvalho (@_felipe) 's Twitter Profile Photo

Copilot auto-complete on Neovim, world leader in plugin deliveries is much less noisier than on VS Code. I now suspect this is why many people hate the constant flow of suggestions and prefer to just prompt on the side bar.

Felipe O. Carvalho (@_felipe) 's Twitter Profile Photo

What people are getting from these AI coding agents feels very close to what people that customized their editor to automate a bunch of tasks for years and years. It's not the time saved! But the good feelings that you get out of having agency over the source-code in a project.

Felipe O. Carvalho (@_felipe) 's Twitter Profile Photo

None of that is extremely slow. The slowness comes from the sheer amount of code generated for LLVM to optimize. Some optimization algos are O(n^3) where n is the length of the IR code. And for incremental builds, you have the linker having to go through a huge number of symbols.

Felipe O. Carvalho (@_felipe) 's Twitter Profile Photo

Yet people measure "programmer productivity" in terms of "number of lines of code produced". In so doing they book that number on the wrong side of the ledger: we should always refer to "the number of tokens spent". — Dijkstra, 1983