[chrismorgan]

Technical correction: YouTube, GitHub and GitLab all produce Atom feeds, not RSS. (Nitter has /rss in its URLs but I can’t confirm they’re actually RSS because the feeds seem to be broken, producing HTML “user not found” error pages.)

(If you want to know more of my beef against RSS (both as a technology and as a term for web feeds regardless of actual format): https://hn.algolia.com/?query=chrismorgan+atom+rss&type=comm.... RSS should have been killed off about fifteen years ago in favour of Atom.)

[leephillips]

RSS seems to have taken over as a generic term encompassing both actual RSS and the superior Atom.

[ravenstine]

I'm curious, why do you think it should have been killed off in favor of Atom? Do you mean in the sense of the terminology, or is there something inferior about RSS?

Perhaps both should be killed off in favor of a JSON-based format.

[chrismorgan]

Refer to my earlier comments: Atom is technically superior and more sound, and calling them all “RSS” is harmful by encouraging the inferior.

JSON stuff? Eh, I feel that’s largely a solution looking for a problem. It’s too late to get universal support in clients (feeds aren’t popular enough and too much of the software basically on life support), so you’re always going to need to serve Atom or RSS as well, and once you’re implementing one of them, why would you bother with the JSON format at all? As for the one serious attempt, JSON Feed, I have one thing that I really like about it and one thing that I don’t. The don’t: it specifies that titles are plain text, which I think is a shame; Atom lets you put HTML in your titles so you can do things like <code>, <em>, <sup> and so forth (and I do, oh yes! I do)—though I wish Atom had let you specify both plain text and HTML (sometimes I want <code></code> to “degrade” to backticks; refer to the titles of my last few blog posts for examples, with plain text in the <title> and HTML in the body). And that’s the thing I think JSON Feed got right for the content, that you can provide both HTML and text, whereas Atom forces you to choose.

[ravenstine]

> JSON stuff? Eh, I feel that’s largely a solution looking for a problem.

Before I say anything else, your viewpoint is totally valid. :) I can't truly argue against it in an objective sense, I think.

It's funny that you say that, however, because that's my perspective on XML in general; it's a solution looking for a problem. Sure, we have Atom, it's here, let's just use it, and I kind of agree. Yet if Atom has already been in a sort of decline, if you will, the complexity it adds only detracts from any sort of resurgence in adoption. The entire syntax is designed to support types, which as far as I've seen is something that has never really been used. I mean, whose feed actually uses custom XML namespaces/tags? Feeds are these verbose and overly-abstract markup files for a purpose nobody really asked for. If someone needs to stick more metadata into their feed, they can already do it with JSON by just adding a non-standard key to the object. Done. No need to have anything beyond primitive types or declaring namespaces.

As far as the HTML thing goes, that's a fair point, but I have to disagree. I don't want HTML to taint titles of things in a feed reader. It's bad enough that people can use emojis and different stylized unicode characters in them, but HTML would just make things worse. I don't need titles to have bold or italics or colors or whatnot. If that's going to be in the actual content body, then so be it, but what you are suggesting sounds like a horrible idea. Granted, I didn't have that problem back when I used to use a feed reader.

[thaumasiotes]

> calling them all “RSS” is harmful by encouraging the inferior.

Where did this come from? You refer to an atom feed as RSS, someone decides to add RSS to their site, and they implement atom because that's how you do RSS. What's the problem?

[chrismorgan]

> they implement atom because that's how you do RSS

No, they implement RSS because that’s all they’ve heard of.

[theshrike79]

> Perhaps both should be killed off in favor of a JSON-based format.

As long as XML has DTD and Schema, it will be superior to JSON