Unfortunately no CSS chapter (again) 😢 We tried in various ways to get the statistics to tell _any_ kind of story but we couldn't get the numbers in time even though several people tried. Even running the queries from previous years proved troublesome. Such a loss for this amazing initiative.

Thought: a new CSS property `all-custom`, similar to the existing `all` property, so we can revert all custom properties on an element at once. Seeing some things in the wild where people manually unset/revert hundreds of custom properties, bloating CSS size.

Quite pleased with myself that I now have the data to support some of these excellent discussions! 😆

Just completed the PR to migrate css-analyzer to Wallace's own css-parser library while maintaining backwards compatibility. Now every Wallace library has a PR to run on the new parser instead of csstree. Now trying to gain confidence to pull the trigger 😬

TIL the deprecated `/deep/` selector combinator At first I thought it was a glitch in my CSS analysis of several thousand websites, but then I looked it up and it was/is an actual thing. web.dev/articles/sha...

Gecko: Intent to Prototype: "Customizable Select"

Started writing a crawler. The easy part. Now need to figure out how to do analysis at scale. SQLite locally with some SQL? Specialized DB for data analysis? Or plaintext files and looping over numbers? I dunno, a bit out my comfort zone here.

Not taking days off between Christmas and New year means you get get to watch everyone catch up on reviewing PR's and messages. Time for another coffee then.

The dust of decades settles on your keyboard as you push aside lines of vendor prefixes Your fedora casts shadows over cryptic CSS hacks from another era. "This belongs in a museum" you mutter ReliCSS: The front-end archaeologist's tool for navigating legacy CSS www.alwaystwisted.com/relicss/ 🙏🖤

My current phone is a piece of crap. Looking to buy a gen 6 Fairphone but battery capacity is lower than my current and reviews are mixed. Any other options out there for clean android with a little more sustainability included?

Quite some lines of code and tests are required to keep this running and I'd very much like to see it go. I honestly don't think anyone ever sees it, and even if they did they wouldn't care about it.

Does anyone have experience using Discogs to catalogue their entire record collection? I've been thinking about archiving my collection for many years and contemplated building a DIY thing, but Discogs seems like a very mature system with so much detail. Seems like a waste to DIY something inferior.

Didn't write a year in review for project wallace this year. I'd always sit down for it because my good friend would urge me and give me pointers with what he'd like to read. My friend died this year of cancer and I haven't even written a monthly update since. Still truly gutted.

Now running in production for a work project! Holding off on merging the PR's I have open for other projects because they involve a lot more moving parts and I 100% want to avoid breaking changes in the libs.

Something that hasn't been made clear: Firefox will have an option to completely disable all AI features. We've been calling it the AI kill switch internally. I'm sure it'll ship with a less murderous name, but that's how seriously and absolutely we're taking this. …

Can I just say that it's such a delight to see @jakearchibald.com handle FF updates and announcements.

Amazing use of free will!

Data folks (esp #rstats) - there’s some untapped data in a BigQuery DB which might need a hand (or two). github.com/HTTPArchive/... It’s for the annual Web Almanac which is a report on the state and health of the Web (incl. sustainability). Probably also nice data for #tidytuesday and #dataviz

ahw, %^$@&%^ it looks like the web almanac chapter for CSS is *really* struggling for completion this year. We're in serious need of some SQL writing capabilities. Without analysis we will not have a chapter to write and we need to skip _another year_.

So I'm supposed to believe that this very uninteresting blog post suddenly has 4 unique visitors from 4 countries in a single day? Need to up my blocking game, these crawlers are out of control. Bounce rates are sky high.

One of my favorite perf hacks: when sorting arrays, create a UintArray of the same length as the input array and sort _that_ one instead of the original. That way you can keep access to the original sorting and have a low-memory sorted one as well.

Was playing with the css parser I'm writing with AI and made some DX improvements. Turns out it's now checks notes 42% SLOWER

it's been 0 days since I've needed this example from @pepelsbey.dev from a couple years ago to complete a task.

Incredible that one of the most powerful takeaways of CSS Day 2025 for me was to use the dryer less often. There was no need for @hidde.blog to hit me that hard.

Cß

I'm looking for my next opportunity in the front-end space 🌐 I've worked at some great companies: Hugging Face, V7, Appwrite, and Significa. I've built NPM packages with 400k downloads/week, and am an avid open-source contributor. Let me know if you want to work together!

This week's Web Weekly is just about to hit all the inboxes! 🥂 If you care about the web platform you should check it out! www.stefanjudis.com/blog/web-wee...

Converted the first library to use this. It need a bunch more convenience methods, but it works!

Not really into AI but found a nice use case: I let it write a CSS parser that is optimized for static analysis. Currently ~3x faster than csstree with up to 52% reduced memory usage; 1.5x faster than postcss (which skips parsing values) and up to 66% less memory with outliers that match theirs

In less than two weeks search engines have picked up this post and traffic is coming in slowly. That wasn't the goal: I just want you to be aware of this option, but it's a very nice side effect.

Trying out syntax highlighting in the terminal for the css-code-coverage package. Not too sure, is it worth it or should I just give up and render plain white text instead?

Gave this a shot for @projectwallace.com: - 🚂 Svelte kit build time: 9.1s -> 5.3s - ⚠️ Build warnings: 0 -> lots, because lightningcss doesn't know `::highlight(AtRule)` - 💾 Page bundle download size: identical - ❌ 1 page erroring completely, seemingly because of css-tree imports somewhere on my end

Does anyone have good experiences with setting up a "break the glass" procedure? Looking to write down some stuff in case something happens so my loved ones can take care of stuff. How do you deal with domain names, github accounts, netlify, bluesky, etc?

Rewrote @projectwallace.com's core css-analyzer package to TypeScript this weekend. Long overdue, but exported types are way better now. It also opens up the door to migrate from css-tree to @eslint.org's css-tree fork which solves numerous issues. And then time for a full rewrite 😈

Yay, took almost half off of the CLI bundle size for css-code-coverage package 🥳 The cli.js now references the core.js in the dist file instead of shipping it's own copy of the core inside cli.js. Surprisingly tricky to fix even though it was basically 2 lines of code change and 1 new line.

CLI question: if you have an argument like `--min-coverage`, would you rather set it as a number between 0-1 or as a percentage between 0-100?

Okay, I'm part of the big cursor club now.

Use your existing Playwright tests to calculate your CSS line coverage www.projectwallace.com/blog/how-to-...

if you create a web scraper/bot without a custom user agent i wish very scary viruses on your computer

How to calculate CSS code coverage with @playwright.dev www.projectwallace.com/blog/how-to-...

Luckily @netlify.com offers pretty fine-grained firewall support, so I've blocked most of the traffic for now. Let's see if it holds up. docs.netlify.com/manage/secur...

Some 600+ visitors (bots, I think?) from Singapore are blowing through my Netlify credits real fast. 100% bounce ratio, weird old pages visited.

In terms of accessibility, would it be acceptable to replace CMD/CTRL+O on a page where the main starting point requires you to select files? On this page I'd expect CMD+O to open a file chooser that lets me select files to start the inspector.

Can we have better CSS Coverage ranges, please? www.projectwallace.com/blog/better-...

Good push to let me write a blog post tonight about a feature that's a bit rough, even though it's not a web standard? I don't even know, I just want it to work better for me.

Very tempted to move all of @projectwallace.com libraries to @eslint.org's fork of csstree. It'll mostly give better nesting support which will solve a bunch of issues across several repositories. And types included.

I need to build+ship an npm package as a library and a cli tool in one package. How on earth do we do that these days? 😵‍💫

Unit tests fails. Grab coffee. Run test again. Still fails.

Open sourcing css-code-coverage github.com/projectwalla... It is *very* early stages, it needs a ton more tests, docs, cli tool etc. but is for the folks who want an early look.