It occured to me the other day that now that the application model for modern computing has largely shifted to the web, and therefore browsers, I find that I often don’t know what to call the silly, just-for-fun one-off projects that I build sometimes.

Take, for example, venmo-calculator - Is this an app? It is a program? It’s technically a compiled golang executable for the backend and a static html+js website that’s compiled in the sense that vue turned a vue component into javascript, but that javascript is itself interpreted?

And yet, when I think of an app, I think of a mobile app - venmo-calculator isn’t such an app, although it could be, if I shipped a PWA manifest. At the same time, something like Windows or MacOS is hard to describe as an app - but MS Paint or the snipping tool, despite not being targeted to mobile, could be described as apps? I could go on about this topic, and I might do that later…

And yet, I know it when I see it

There is, however, one category of programs/executables/applications/apps that I can clearly delineate - in the words of Potter Stewart - I know it when I see it - command line tooling. The purpose of this post (surprise!) is to discuss the different command line tools that I use frequently, and why I like them.

• rg: ripgrep is a grep-like search tool that is extremely fast. Part of the magic is that it happens to be written in Rust, and also respects .gitignore files (you can optionally turn this off, of course). I picked up this tool at Spokeo because a pretty common task we had to do was find a particular row in a very large CSV file that was malformed, to fix it. A few also used Silver Searcher (ag) - but in benchmarks rg was slightly faster, and these days ag is not as actively maintained.
• nvim: neovim is a fork of vim that’s easier to work with when it comes to plugins and other tweakables. Also, uses python for programmability, which is nice.
• direnv: direnv lets you manage your environment at the directory level using .envrc files, and you can direct those files to inherit from other files, too. It’s not just environment variables - just like .bashrc it executes whatever is in the .envrc file, so you can use it to run prescripts or other tooling.

Quite a few of my favorite tools happen to programming language version managers - whenever I initialize a new system, I always start with these instead of directly installing programming languages

• nvm: Node Version Manager - for node.js
• pyenv: pyenv - for python
• gvm: gvm - for golang
• g: g - also for golang. I slightly prefer g since its less verbose, and also does a good job of dealing with self-dependencies (in newer versions of golang you actually need golang to compile)

I also use a few different meta-installers for tools that use language-centric installers:

• pipx: pipx is an installer for python tools that provisions an isolated environment for each tool - no more dependency conflicts just because of your tooling! This has become my tool of choice for installing python-based linters and dev tooling such as pre-commit and sqlfluff
• uv: uv is a package installer and resolver written in Rust that is super fast. It’s meant to be a drop-in replacement for pip - I can attest that it’s faster, but I don’t always remember to use it. Also, it’s a resolver and not a runtime environment, so complex projects benefit the most from the speed, whereas libraries that have long and unoptimized install scripts don’t really benefit (I’m looking at you, tensorflow).

Python-related stuff:

• bpython - bpython is a fancy auto-complete and syntax highlighting layer on top of the standard Python IDLE. I don’t often use the python interpreter directly much anymore (much more common at Spokeo, these days I like to use the Jupyter integration in VSCode), but I try to use bpython if I have to (and if I can remember).
• polars - polars is a python API to a vectorized data processing engine written in Rust - so it is FAST! It is also, in my opinion, more ergonomic than pandas - and it doesn’t have heavy baggage of broad usage, yet. I do think that the baggage of pandas is due to its lineage in R, though.

Other terminal-related tooling:

• starship - a very fast and customizable prompt for most shell environments
• warp - my preferred terminal app for mac and linux (on windows, I actually like Windows Terminal). I don’t find the AI features very useful, except for autocomplete. ⚠️ - warp is closed source, unfortunately.
• monaspace - a font family designed by Github that I find pleasing to use in VSCode. There’s also a version that’s had Nerd Font glyphs patched into it