Thread by @lzsthw

Another element of my "why browsers must die" linked list is how some parts of it assume everything is a flexible rectangular viewport with fixed dimensions, and other parts assume it's an infinite plane with no fixed dimensions.

For example, scroll into view:

If a browser is an infinite plane that you view through a viewport (like a video game), then there *must* be a way to cause interactions when some element of the infinite plane comes into view. Yet, finding an event for this very necessary thing is impossible.

Here's a hilarious stackoverflow:

stackoverflow.com/questions/4870…

It was asked *11 years ago*, but was active *11 days ago*. It's been viewed 727k times, and has as many possible answers as there are people posting over various years.

The way to explain this failing is, imagine you're using a game engine, and you build this massive 2D world, like a side-scroller Mario style, but for some weird reason the devs on the engine decided to *not* let you know when one of your characters is visible in the viewport.

When you ask for this feature, they whine back at you, "BUT, this game is all about the viewport. You should be able to work just within that." When you point out that you've built this whole huge side scrolling world like they said you should, they go, "But...uhh PHONES!"

Alright, so if it's all about the viewport, and that's the only thing you're allowed to actually interact with, then why can't we decide that a viewed webpage *only* fits the viewport? If I can't get events when the viewport is moved, then I need to remove infinite scroll.

When you demand that, they go, "OHhh nooooo, you see we're a *document* view system, not a *screen* system." That means, you *have* to accept an infinite scrolled and sized document and can't declare "this document exactly fits the screen".

Ok then...infinite field...

Well then if it's an infinite field that I display to the user with a viewport...then why can't I get an event when an element of the *document* is in view?

Honestly, I'd love these browser companies to stop hiring HTML/CSS gurus and start hiring game engine developers.

So, there is a proposed solution to this called the "intersection observer":

developer.mozilla.org/en-US/docs/Web…

And, it's the typical "we tried to copy Java and failed" code style withAllYourCamelCaseBelongToMe big wording and complex APIs. Crafting objects that call callbacks that call..

Code like this just really irritates me to no end. IntersectionObserver? I have to build my own threshold ratios? Look, all I want is this:

<div onVisible="doThatThing()">

And you figure this junk out. Computer do math for me. Not the other way around.

So once you do all of that, how do you find out if something is visible in the viewport?

"It's easy! You just have to create an observer and then monitor ratios across a 5th dimensional hyper plane using abstract cartesian coordinates mapped to a Hilbert space monad."