utensil.tngl.sh / blog
Utensil's Hugo blog for math, tech and thoughts utensil.tngl.sh/blog/
0
Fork 0

Configure Feed

Select the types of activity you want to include in your feed.

blog / scripts /
at main 8 files
utensil retire(blog): drop scripts/{freeze,setup-freeze-env}.sh + freeze-era docs
00a2990b
README.md
check_all.sh
dev.sh
install_pikchr.sh
install_typst_ts_cli.sh
make_all.sh
make_changed.sh
make_single.sh
README.md

Build scripts#

Diagrams + dev#

  • make_all.sh — render every *.pikchr / *.d2 / *.typ under content/posts via make_single.sh, then check_all.sh asserts the expected SVGs exist.
  • make_changed.sh / dev.sh — incremental rebuild + live preview (watchexec).
  • install_pikchr.sh, install_typst_ts_cli.sh — fetch the diagram binaries.

How the blog is rendered#

Two independent lanes, no cross-deps:

  • github.io lane — .github/workflows/gh-pages.yml does a LIVE render on GitHub Actions (julia-actions/setup-julia + xvfb-run quarto/hugo) and publishes to https://utensil.github.io/blog/.
  • tngl.sh lane — .tangled/workflows/{build-depot,render}.yml runs on tangled's Spindle: build-depot.yml pre-builds the precompiled Julia depot
    • render-env closure (nix flake at repo root) and uploads to Cloudflare R2; render.yml pulls them via manifest.json and renders LIVE under Xvfb, publishing to the site branch → https://utensil.tngl.sh/blog/.

No freeze step is needed any more — the ca-in-julia Julia post executes on both lanes, and _freeze/ is no longer relied upon. Quarto's freeze: auto remains on (in _quarto.yml) for cheap local re-renders only.