A version number that lied by three releases
The Open Source page on this site lists four packages I built and share publicly, each with a little version badge. v0.6.2, v1.0.0, that kind of thing. I typed those numbers in by hand when I built the page. What I hadn’t done was ever go back and update them.
A review pass pointed out the obvious problem: a hardcoded version on a public page is a promise you have to keep manually, and nobody keeps that promise. I went and checked how bad it was. One of the four was showing v0.6.2. The real published version, the one anyone installing it today would actually get, was three minor releases ahead. The badge wasn’t just stale, it was actively wrong, and had been for a while.
The fix wasn’t computing it
It was deleting it. A version number on a marketing page for an open source package answers a question nobody visiting that page is asking. What they want to know is whether the package exists, what it does, and how to install it. None of that needs a digit.
That last part turned into its own small fix. The install instructions on each package page were showing the exact dependency line this site uses internally - a pinned GitHub tarball URL, because that’s how this site happens to consume its own packages. But these are published to npm now, and nobody evaluating whether to use one wants to see this site’s private plumbing. They want the command they’d actually type. So the install box now shows a plain npm install, with a real link to the npm listing next to the GitHub one.
Lesson
“Make it accurate” and “make it useful” aren’t always the same fix. I could have spent time building a mechanism to keep that version number honest. The honest move was admitting the number never needed to be there - the opposite conclusion from the version badge I did go correct a few days earlier, where the whole point of that page was to say what this site runs.
Related reading
The open source page that almost lied about its own version numbers
Four packages, four latest tags, and the thing I should have checked instead - plus a content collection I talked myself out of building.
The filter pills existed, just not where anyone could find them
Per-project filtered views with a nice pill nav - reachable only from homepage cards. The main /blog index, where everyone actually lands, had no way in at all.
The CTA that pointed at the wrong dev log
The only call to action on the Deep Cut Atlas page linked to the whole unfiltered blog - a link that worked, returned 200, and quietly sent everyone to the wrong place for weeks.