I thought about posting notes on Syncthing but then I’d have to tangentially talk about NixOS. NixOS is my main Linux distribution atm but Nix/NixOS/Flakes are too hard to write about and I fear they’ve ventured too far into the realm of over–engineering.

What do I mean by over–engineering? An over–engineered tool is one where even the simplest use cases are non–obvious (to most people) and this can happen when it tries to do too many things with “specificity”. The overall concept is elegant though (explainable in lay terms) and can be applied in other contexts.

Here’s a final muse for current year. This probably counts as wise posting. On local news there’s sentiment that the world is now “post–truth”… but historically; since when has the world not been post–truth? Truth (verifiability/provability) is ultra rare and trust abundant :-)

The Internet/Blockchain/whatever is not a truth machine and could never be a truth machine. Inside events/stories/happenings everyone’s just as clueless as in past eras and technological differences become “functionally equivalent” — meaning you either waste time separating signal from noise in “fast” and “fake” information infinitum or waiting for “slow” and “official” carrier pigeons to arrive.

I’ve since realized that Hugo’s architecture provides a variety of template optimization strategies. Hugo builds pages concurrently, so it might be hard to see on a modern device but before partialCaches or module mount trickery — there’s still the implicit complexity of the output/lookup model.

Generally the complexity cost of the default output formats are: page > term > taxonomy > section > home. Keeping expensive calls inside a section and/or a home template is usually optimal. i/o and maybe memory should be the only problems with lots of pages.

The blog linked in a previous post is a gem. Too bad the current site doesn’t appear to have all the archived posts, you need strong search–fu to find them on archive.org.

Hugo is a gateway for discovering neat golang libraries. Version 0.104.0 introduced a color extraction method that has lots of use–cases. An easy one is to generate basic image gradient placeholders. The browser has its own deferred/lazy loading logic so fancy image gradients (on a static site) require only a few lines of pre–generated CSS styles.

Reverse pagination is a counter–intuitive strategy for attempting to make links immutable/cacheable and bookmark friendly across older pages. I searched for a visual explanation (difficult to explain concisely) and eventually arrived at an old article on paging apis. Reverse pagination has its gotchas, but then again pagination itself is one big gotcha.. :-) Well, it depends on the use case really.

I was reading web.dev recently and couldn’t help but think that in tech circles/articles online it’s easy to get the impression that Firefox is a major competing browser. Firefox actually doesn’t even register. Firefox has an estimated 3% points in global market–share and doesn’t even show up in mobile market–share estimates.

Users implicitly use Safari (Apple) and/or some derivation of Chromium (Google Chrome). When was the last time you saw someone irl using Firefox? Not recently if you’re outside the tech bubble.

It’s kinda neat how CSS animation rules are sort of simple in their construction. imo, rules could be made even simpler if the animation-delay property also allowed delays between iterations/intervals instead of at the start only. Interval delays could allow for writing drastically less key frame rules.

text-animation[hang] {
  animation: tilt-rightward 1.3s infinite, tilt-leftward 1.8s infinite;
}

The temptation to bring in a js bundler is oh so very great. Deno bundle is obviously not designed to bundle js directly for the browser but you can get away with it up to a certain point.