# Roux A slow-publishing fashion journal — one hundred photographs at a time. Editorial, couture, beauty and backstage plates. **Live:** [roux.pivoine.art](https://roux.pivoine.art) --- ## Stack | Layer | Tool | |---|---| | Static site generator | Hugo Extended | | CSS | Tailwind CSS v4 | | JavaScript | Vanilla (no framework) | | Package manager | pnpm | | Server | nginx (Docker) | ## Development ```bash pnpm install pnpm dev # Tailwind watch + Hugo dev server (http://localhost:1313) ``` ## Build ```bash pnpm build # compile CSS → hugo --minify → public/ ``` The compiled CSS at `static/css/main.css` is a build artefact and gitignored. Always run `pnpm build:css` (or `pnpm build`) before committing changes that affect Tailwind output. ## Content Posts live in `content/posts//` as Hugo page bundles. Each bundle contains: - `index.md` — frontmatter with `plate`, `title`, `description`, `categories`, `tags`, `issues`, optional `featured: true` - `.png` — the plate image (used for thumbnails, lightbox, og:image) Issues are defined in `content/issues//_index.md`. ### Importing posts Content is sourced from `~/projects/ginger/posts.csv` and matching images in `~/projects/ginger/images/final/selected/`: ```bash python3 scripts/import-posts.py ``` This creates or updates page bundles, assigns plate numbers (sorted alphabetically by filename for stability), normalises descriptions, and copies images. ## Design tokens All colours, fonts, and spacing are defined in `assets/css/main.css` inside `@theme {}` and mirrored as CSS custom properties in `:root`. Edit there — not in a Tailwind config file (v4 has none). Key tokens: `--color-paper` `--color-ink` `--color-roux` `--font-display` `--font-serif` `--font-sans` `--pad` `--gap` ## Deployment The repo includes `nginx.conf` and a `Dockerfile` for containerised deployment. Build the image and serve `public/` via nginx. Analytics (Umami) are injected only when `HUGO_ENVIRONMENT=production`. ## License All editorial content, photographs, and design © Roux MMXXVI. All rights reserved. Contents are explicitly excluded from AI/ML training use without written consent.