Oddly enough the RSS spec has an optional comments element. I wonder if there’s anyone using that in an interesting way.

If reader mode fully takes off, then there’s no need to worry about styles in specific situations (articles). Offer some bare minimum semantic HTML and you’re good. Firefox’s reader mode wins hands down right now. Content before context.

Sometimes it’s amusing to stumble upon the latest centralization/decentralization debate. You could argue technically against centralization, spof, and censorship until you’re blue in the face.

Decentralization is pretty much a solved technical problem, but for very special non–technical reasons everything only gets more centralized from this point onwards :-)

Parsing RSS feeds is always a pain. Don’t parse feeds inside the template engine kids :-)

SVG is a solid tactic. No one “inspect elements” every site. Everyone’s a user, even programmers — they won’t know until someone tells them.

I just noticed that Mastodon prefixes a user’s homepage URL route with the @at symbol. Pragmatic and clever name spacing? Tempting. Though I’d still prefer a URL of the form:

https://example.com/user

Instead of:

https://example.com/@user

Yeah… superficial, but it looks cleaner.

A lot of websites dump the entire feed — that’s nice. Ideally for bandwidth efficiency, combine multiple delivery strategies.

1. A partial content feed (Atom preferred) limited to the last 3/5/10 posts for quick queries.

2. A full content RSS feed of a similar limit for fresh content.

3. A hidden/unlisted feed for dumping all content (mirror).

Alternatively, just implement atom’s collection partial lists for feed pagination. If your post bandwidth (size/frequency) is low, then pretend you didn’t see this — be lazy and optimize later.

Here’s a great article that implements reading webmentions natively in hugo. There’s always a soul out there who has tried what you’re thinking about.

Web mentions and reply by email? The first is easy thanks to webmention.io, but the second is very involved and requires a public email inbox setup.

Fun fact: fetching favicons from various domains is a dark art. Might seem trivial at first, but it’s very complicated. A simple request for a favicon.ico will net strange returns — like serving a text file as a favicon.ico.

Better to leverage a browser which excels at interpreting favicons or contact a CDN of a major search engine. Make sure to cache the results.

• https://icons.duckduckgo.com/ip1/en.wikipedia.org.ico
• https://www.google.com/s2/favicons?domain=en.wikipedia.org