The content store, part 1: stop fetching what you already have
Why we stopped fetching from our CMS on every request and started syncing everything into a local SQLite database instead. Part 1 of a series.
psousa.dev=#selectbiofromauthor;
I build the content infrastructure behind large multilingual websites, and tools for data that outgrew its container.
select*frompostswherefeatured= true;
Recreating macOS Magnet window management in Linux
Set up macOS Magnet-like window management in Linux using gTile with custom presets and keyboard shortcuts for efficient workspace organization
SSR CSS modules as a library author
Rollup-compiled CSS modules add a client-side script, raising concerns for server-side rendering. Navigate this quirk for seamless integration in both environments
select*frompostsorder bypub_datedesc;
A URL redirector that never watches you
Most URL redirectors work by reading every request. Ferry pushes the rules down to the browser with declarativeNetRequest, so it redirects without ever seeing where you go.
Building a JSON viewer for large payloads
JSON viewers are easy to build until payloads get large. JSON Bonsai focuses on faster navigation, search, queries, and data fidelity.
The content store, part 4: the fun stuff
Once your CMS content lives in a real database, weird doors open: ranking components by fold position, generating your imports from usage data, and CI builds that never sync. Final part of the series.
The content store, part 3: the read side
Serving CMS content from a local database sounds trivial until stories start referencing each other. Relations, cycles, and why dev mode reads differently than production. Part 3 of a series.
The content store, part 2: keeping a copy you can trust
Syncing content into SQLite is easy. Keeping that copy honest is the hard part: redundant writes, locales that disagree, and syncs that fail halfway. Part 2 of a series.
select*fromprojects;
URL redirector that never watches you. Native declarativeNetRequest redirects, no telemetry.
Terminal radio for wandering the world's airwaves. Privacy-first, offline search, no telemetry.
Browser extension that keeps 100k-node JSON payloads navigable. Tree, table, schema, and query views.
Local npm package development without the reinstall loop. A faster yalc.