Paper Weight

Hugo: A Solutions-First Web Stack

· JSON Alexander

I spent the last week rebuilding parts of my site stack and figuring out what I actually want my workflow to look like going into the next few months.

Turns out the answer is probably:

  • static sites
  • Go binaries
  • cheap infrastructure
  • terminal-first tooling
  • and significantly less JavaScript than the modern internet would like me to use ( but not no javascript ;) )

Getting Going

Hugo felt opaque when I first fiddled around with a theme.

You make a post, things render in an unexpected way, themes inject layouts (really weird compared to what I’m used to) from far away, and tracing partials through themes felt like playing Where’s Waldo inside a filesystem.

On day three of messing around with a few themes however, I started to get it. The moment I stopped treating Hugo like a frontend framework and started treating it like a compiler, the architecture suddenly made sense.

But here’s the deal - Content lives in content/ - Layouts define rendering - Markdwon gets compiled into a static artifact - There is no runtime, db, forest of dependencies

And that’s why its awesome for bloat-hating enterprises and indie devs alike.

By The Way

A lot of this is also downstream from moving more of my workflow into Linux full-time.

Kitty + tmux + neovim + Hugo + GitHub Actions feels absurdly fast compared to bouncing between GUIs for everything.

I still use Windows for Unreal and some school stuff, but for building software I increasingly just want composable tools that do one thing well.

The current setup is roughly:

- Arch Linux
- Neovim
- Hugo
- GitHub Actions
- S3 + CloudFront deployment
- Go-based tooling where possible
- Bash scripts for glue

Which also explains why I’m becoming increasingly obsessed with Go binaries.

Build Queue

The next few months are probably going to be extremely build-heavy before I start working full-time.

Current queue looks something like this:

Farevoyant

CLI-focused travel/project system built around Cobra with a Hugo frontend.

Still figuring out exact scope, but I want something terminal-first with static output and minimal infrastructure requirements.

FOSS Go Binary + Release Pipeline

I want at least one polished open-source binary with:

  • goreleaser
  • GitHub Actions
  • automated builds
  • versioned releases
  • cross-platform support
  • proper documentation

Mostly because I think distributing software should feel clean.

Site Infrastructure

A few different sites are in progress:

  • Coda (business/project hub)
  • Curriculum Vitae
  • Paper Weight
  • A more design-heavy experimental site
  • Potential landing pages for family projects
  • Game project sites
  • National park trip logs

Most of these are excuses to experiment with deployment architecture, design systems, and static infrastructure at scale.

Market Analytics Database

Still undecided on exact direction, but likely:

  • relational storage
  • ingestion pipelines
  • financial datasets
  • analytics tooling
  • portfolio simulations

Possibly overengineered. Hopefully educational.

Portfolio Backtesting Engine

I want to build this mostly because I think too many finance tools are black boxes.

I care less about pretending to be a quant and more about understanding:

  • simulation architecture
  • data flow
  • reproducibility
  • performance tradeoffs
  • visualization

Bash Scripts

Honestly just a growing pile of utilities for:

  • deployments
  • media conversion
  • filesystem cleanup
  • backups
  • server management
  • homelab maintenance

Half of Linux is eventually realizing you should’ve written a script three weeks ago.

Unreal / Horror Prototype

This one is simultaneously the least practical and the one I probably care about the most.

The current goal is simple:

Build a short, visually convincing horror experience with strong environmental audio, grounded lighting, and enough production quality that it feels real at a glance.

I’m intentionally trying to keep scope under control:

  • small environments
  • linear progression
  • modular assets
  • baked lighting where possible
  • atmosphere over mechanics

We’ll see how long that lasts.

Anyway, that’s the current direction.

Less “learning web development.”

More:

  • systems
  • tooling
  • infrastructure
  • deployment
  • pipelines
  • and building things that actually ship.