Wait, the Document Outline Algorithm is dead? Why? Who the fuck enjoys manually choosing an appropriate <h1> - <h6> element when the structure could be inferred from the container??
After all this time, it's rare for me to use CSS and get confused by its behavior, but today was such a day.
I had an element with "position: relative" and "&::after { position: absolute }". Whenever the ::after element was displayed, the outer element's size changed, but only if it also contained a child element with a margin.
This shouldn't happen? Why would an absolute-positioned element affect anything? I chose "::after { position: absolute }" specifically to add a focus indicator without it interfering with the layout.
So my boss hates io-ts and I thought I could use regular TypeScript interfaces to keep frontend and backend route types synchronized. It would work, if it weren't for Date objects and being automatically sent as strings.
@privateger There are so many ways this could be solved. Set the cache to 5 min so it handles large bursts of requests. Load a cached article and use JS to swap cached ads with dynamic ones. Look at the user agent and use a cached version if it matches Mastodon (and other known scrapers), if you insist on dynamic ads without JS. That took me 2 minutes of brainstorming, like come on.
Volpeon 
