Kevin Cojean

Tired of Python for scripting... going to give Scala a chance for scripts

Articles

  1. Be a thoughful and nice developer
  2. Speech to text for Opencode
  3. Flight migraines
  4. The simplest morning routine which gives me the...
  5. My Taskwarrior Workflow
  6. Tiling managers for KDE Plasma on Ubuntu
  7. [Vim] You should use commit hashes in your vim-...

Code snippets

  1. Setting up a backup AI provider in case Opencod...
  2. Getting shift+enter to work with Kitty with Tmux
  3. Nuke the LinkedIn feed
  4. Taskwarrior TUI - A nice theme override
  5. Using different Git identities for different pr...
  6. Your favorite dev tools shouldn't pollute a sha...
  7. Combining the asdf version manager with the Jav...
  8. Automatically snoozing all incoming emails unti...
  9. Why Google Antigravity beats Claude's subscript...
  10. LLM auto-complete plugin for Vim for OpenAI, Mi...
  11. [KDE] CTRL+T Shortcut: Focus or Launch Konsole ...
  12. [Vim] Proper code folding for Python files
  13. [Vim] How to compile Vim
  14. Researching the best small VPS for under 5 euro...
  15. [Vim] Fuzzy finding a string within a project w...
  16. [Vim] Prank your coworker by forcing them to us...
  17. A monadic Result type using generics in Python
  18. Interacting with Bitwarden in your terminal
  19. [Vim] Disabling typehint diagnostics for the `p...
  20. Free git repository with actual large file stor...
  21. Script utilitaire pour générer automatiquement ...
  22. [Vim] Navigating between Tmux panes using Vim m...
  23. Autogenerating an OpenAPI json file using Maven

Thoughts

Next Previous

Tired of Python for scripting... going to give Scala a chance for scripts

Probably you should remind before EACH message to a LLM to apply RED/GREEN testing... it's really important they don't "forget" this. And it helps a ton reduce token and time waste

Trying LazyVim was a mistake. I'm just fighting it's auto-magic configurations all the time...

Avoid vendor lock-in at all costs, this was true for IDE's, this is more true for your AI harnesses.

I am loving my decision of centralizing all my LLM requests through a llm-proxy it's proving very helpful now that lots of services are struggling to scale with demand

If the hardware supply continues to balloon in pricing, do you think there will be a divide between programmers who can code manually (cheaply) and those who never learned how to not-vibe-code?

minimax-m2.7 seems incapable of tool calling ?

For configuration variables: flat .properties files > .json > .yml

Dependency injection is the best. Don't let anyone tell you it's not pyThoNiC or something...

Turns out, LibreOffice on an AppImage is SO MUCH snappier than using apt or Flatpak installations; a relief!