Ask HN: Which tools have made you a much better programmer?

probably you will like

  function fixup() {
    git commit --fixup=$1
  }
  function refixup() {
    git rebase -i --autosquash --autostash $1^1
  }

I moved to awesomewm a few years back and I absolutely love it. I can open half a dozen terminals at once and they'll all automatically be ordered in a way that's immediately useful to me, where all the terminals are visible and (largely, depending on the automatic layout set) equally sized. And if I want another layout, I just press one combination to cycle through all the layouts I've configured. tmux doesn't give me that kind of flexibility. It doesn't feel anywhere as fluid or seamless to switch between half a dozen (or more) terminals.

It means I can have an overview of a bunch of different things and keep terminals context-specific (1 terminal for htop, 1 for docker, 1 for whatever remote test environment, 1 for project A, 1 for project B, 1 for some other remote host I need for some reason, etc.) If I want to do a new task unrelated to anything I'm doing before, I don't need to break the context of an existing terminal, I just press Alt+Enter, it's automatically slotted into a place where it's completely visible and usable and I can do that task quickly. When I'm done, I can close it, again, without disturbing the context of all the other terminals. It's just incredibly freeing to have that and I feel it frees a lot of cognitive load by being able to go back to a terminal for a certain task and immediately see exactly where I was and what I did last.

Also, much like the other comments, I use task-specific virtual desktops all the time. First desktop is for all the terminals. Second is for browser/communication. Third is for project A. Fourth is documentation related to projA. Fifth is projB. Sixth can be more documentation. I often have 10 virtual desktops for different things. I don't want to imagine what it'd look like if I had it all on one desktop.

The static analyzer. There are so many subtle things that aren't caught by the compiler and don't cause any problems in your testing but will definitely come up once 10M users are running your app. The static analyzer can catch some of them before you even run! It's great.

The various sanitizers - Address Sanitizer, Thread Sanitizer, Undefined Behavior Sanitizer. These all also find things before you ship. They require (re)building and running but it's worth it.

Beyond that, any sort of instrumentation that can show you in real time information about your running application. You can see memory growing, even if it's not leaking. You can see performance tanking and know the exact function that's causing it before you quit.

Beyond that, for me using a framework that was well written made understanding architecture much easier. I had used MFC and CodeWarrior PowerPlant for a few years. They got the job done and were easy enough to use, but I didn't really learn anything from them. But when I moved to using Cocoa, it was so elegant and the separation of concerns was so good that it clicked and really taught me a lot about better architecture.

Not strictly a better programmer but a more efficient one.

- FZF (Fuzzy Finder) really speeds thing up in bash when searching your history.

- Vim (as others have mentioned). With the plugin vim-fugitive (a git related vim plugin). I love Gblame in vim-fugitive, as being able to look at the history of lines of code and stepping back in time can at times be very useful.

- Creating various aliases. For example in git I have little aliases which just make me faster. "st" = "status", "rbi" = "rebase --interactive", "ci' = "commit", etc...

- Using The Silver Searcher (ag) or Ripgrep (rg) instead of grep. Much faster than using grep.

- Using 'fd' instead of find. Like above it is a lot faster.

Interesting. How does that work when you want to rest your forearms while typing?