[berkes]

I've never really liked JSON as replacement for XML. Had we continued with RSS, Atom and XML, not only would we make peer to peer distribution far easier, we'd have a very easy publishing mechanism.

But we threw out XML for JSON. With JSON we need loads of custom, client side code to turn it into a DOM that the user can look at. With XML we only need XSLT. It won't work for all cases, but the majority of sites wouldn't need a single line of JS to renders sites. Yet here we are: shadow DOM, event listeners, useeffect, JSX, progressive hydration, and so forth and so on. To build web-experiences that we could deliver back in early 2000 but were deemed too complex and too daunting.

[michaelmior]

To be fair, many of the things you listed are able to implement features not possible with XSLT.

[berkes]

Not sure if I understand correct, but the things I listed (event listeners, AJAX, etc) are things needed to be implemented so that you get what XML+XSLT in a browser give you for free.

To be clear: there are lots of things that XML+XSLT cannot do, but which JavaScript+HTML can do (in an HTTP context).

But for the most typical HTTP usecase: a website presenting information, XML+XSLT is fully up to the task, yet we forego that, and instead pull out the big, complex "guns", entire JS frameworks. A knee-jerk reaction that I blame on the bad rep XML got, and the praise it's incomplete but simpler replacement, JSON, got.

[ssdspoimdsjvv]

Good news, you can nowadays transform JSON using XSLT, even in the browser.

[kmac_]

JSON is great, but lacks one feature that XML and its ecosystem has - extensibility and deep standarisation. XSLT, XML Schema, XML signing and encryption, native support of ids and refs, etc. All that missing points are doable with JSON or were added quite late, but yeah, standarisation is important.

[pjerem]

JSON somehow have standardisation with OpenAPI.

But yeah, JSON is a pain. XML was pretty smart.