bendichter.com

Personal site, built with Astro. This is a prototype replacement for the AcademicPages Jekyll fork the site ran on previously.

Running it

npm install
npm run dev        # http://localhost:4321
npm run build      # writes ./dist
npm run preview    # serve ./dist

Deployment is a GitHub Actions workflow (.github/workflows/deploy.yml) that runs npm ci && npm run build and publishes dist to GitHub Pages.

Where things live

Path	What it is
src/content/posts/	Blog posts, markdown with front matter
src/content/talks/	Talks, markdown with front matter
src/content.config.ts	Front matter schemas, enforced at build time
src/lib/content.ts	Cross-file content checks (dates, duplicate URLs)
src/data/site.ts	Name, nav, social links, the software list
src/data/publications.json	Generated from ORCID, do not edit by hand
src/data/publications-extra.json	Papers missing from ORCID, edited by hand
src/assets/	Images referenced by posts, optimized at build
public/	Files served as-is: favicons, CV, .well-known

Adding a post

Drop a markdown file in src/content/posts/ named YYYY-MM-DD-slug.md:

---
title: "Post title"
date: 2026-07-30
permalink: /posts/2026-07-30-slug
tags:
  - python
---

title, date, and permalink are required and the build fails without them. Reference images with a relative path (../../assets/thing.png) so Astro optimizes them; a path that points at a file which does not exist is a build error rather than a broken image on the live site.

Updating publications

npm run publications

This rewrites src/data/publications.json from the public ORCID API, dropping the per-release software DOIs and collapsing preprints into their published versions. Commit the result, so builds stay offline and reproducible.

ORCID is currently missing some papers, so publications-extra.json holds those by hand. Every entry in that file is a gap in the ORCID record. Adding the work to ORCID and deleting the entry here is the better fix, because ORCID feeds everywhere else too, not just this site.

Why the build is strict

The schemas and checks exist because the previous site shipped all of these to production and nobody noticed for years:

• A talk dated 2012 in its front matter and 2020 in its filename, which sorted it below a 2018 talk and made the talks page look abandoned.
• A post permalink reading stacked-stop-plot instead of stacked-step-plot.
• A publication link that was two URLs concatenated by a bad markdown conversion and resolved to nothing.
• The same paper listed twice, as a preprint and as the published version.
• Thirteen favicon files referenced by the page head but never generated.

Making each of these a build failure is cheaper than finding them later.