tmux VS neovim

Instead of opening a bunch of terminal tabs or windows, I switched to tmux. It lets me manage multiple sessions in one window, split panes, and run different services side by side. It’s lightweight, keyboard-driven, and fits perfectly with NeoVim.

If you've used terminal multiplexer in command line, you know tmux is cool! If you haven't, you really should use something like tmux, especially if you SSH into remote servers often!

Additionally, I integrate several CLI tools into my work flow, such as lazygit for streamlined Git operations, yazi as a terminal file manager, tmux for session management, and lazydocker for handling Docker containers efficiently.

3. Running the App Entirely in Docker (with Persistent Data): For devs who prefer a fully containerized development environment, they can now the backend and database in Docker (my personal favorite method). This approach minimizes dependency conflicts and leverages Docker-specific PostgreSQL tools. To ensure persistent data storage, similar to a locally hosted PostgreSQL database, I configured Docker volumes. With Docker volumes, this enabled both staff developers and contributors to fully containerize the application without needing to re-populate the database with each new container. Additionally, this streamlined my pull request workflow as a maintainer, as I no longer needed to manually populate the database from a forked branch when reviewing complex pull requests locally. Of course, there are caveats to this method, forked pull request tests run on my machine using Docker volumes can alter my local database, but I quickly realized I could navigate this using tmux multiplexers or docker-compose.override.yml files (that is for a future blog post).

When systemd broke tmux (which isn't a Linux project, but ported from OpenBSD) and instead of reverting or fixing their own bug, systemd devs went to the OpenBSD folks and asked them to work around the bug that they caused themselves. This is ragebait-level insolence:

https://github.com/tmux/tmux/issues/428

This. To add some words why this is important:

Given the remote-first container-based world we're heading towards, decoupling UI (terminal emulator) from its state (tmux, code-server) is a great design decision, which I think will ultimately define what the "next generation" of terminal emulators is. Imagine being able to open tabs directly on remote host, reconnect without losing state, etc, all while using native UI (so Cmd+T to open new tab, Cmd+F to search, etc). Productivity game changer, which currently only the iTerm2 users can fully enjoy.

Ptyxis (putting its state in running containers), WezTerm (native handling of ssh sessions) and VSCode's terminal (starting a proprietary code-server binary and connecting to its TCP port) have reached some of this functionality, but in their design they need some out-of-band mechanisms to do their magic, ultimately limiting the scenarios they can handle.

Meanwhile tmux -CC [0] and ht [1] are sending both their control channel and data channel over the opened terminal itself (in-band), making them flexible enough to support any configuration. Something complex like `ssh jumpbox -- ssh prod -- podman exec -it prod /bin/bash -- tmux -CC` should just work.

[0] https://github.com/tmux/tmux/wiki/Control-Mode

[1] https://github.com/andyk/ht

What is tmux?

This also gave me the chance to learn how to use Tmux. Best tool I've learned in a while.

To keep it running in the background we can use tmux

Vim is a highly efficient, keyboard-driven text editor that prioritizes speed, precision, and control. It’s been around for decades, but it’s far from outdated. In recent years, Vim has seen a resurgence, thanks largely to Neovim, a modern refactor of Vim that brings a faster core, better plugin support, and a vibrant, forward-looking community.

A text-based shell interface is extremely lightweight compared to modern IDEs. Because it runs in the terminal, even a feature-rich agent has very low overhead. According to Forgecode “Low Resource Usage: minimal impact on system performance”. In contrast, a full IDE can consume hundreds of megabytes of RAM or more, even when idle. In one user benchmark, Neovim (a terminal editor) used only about 10 MB of RAM, whereas Visual Studio Code (an Electron-based IDE) used roughly 700 MB with no files open. The savings add up quickly: even a hundred developers using shell agents could free up many gigabytes of memory compared to the same number running heavy IDE instances. In practice, a shell agent like Forge leaves almost all CPU and RAM free for your code compilation and tests. In a cloud or CI/CD pipeline, this efficiency translates directly into cost savings. You can run more parallel analyses or smaller instances when the tools are light. Over time, those saved resources mean lower infrastructure bills for large teams.

So I headed over to the Neovim GitHub like any lost soul would and followed the official install steps:

Neovim GitHub Repository

PDE (Personalized Development Environment): Neovim

I've used VSCode for years. It's great, no doubt. But when you have to juggle up to 5 projects daily, you'll find that it's going to block you away. Mine feels heavy, especially with multiple windows and extensions. This is why I decided to find an alternative, until I started using NeoVim instead.

As the developer of the Automad flat-file content management system, I wanted to make working with Automad themes in Neovim as smooth as possible. Automad has its own templating syntax, so I created tree-sitter-automad to provide proper syntax highlighting. Since it’s not yet an official parser, you need to manually register it in Neovim. To speed up template writing, you can also add custom snippets using LuaSnip. Here's how to set everything up.