you could still have your html in a template file and parse and return the contents of that.
On first read, I thought by "generating html from function calls" the author was referring the react-style way of building a UI by writing a reactive function that returns HTML.
Their reasoning: https://firefox-source-docs.mozilla.org/devtools/frontend/re...
As an example, here's the Editor main component (same "latest" commit you pointed to):
https://github.com/mozilla/gecko-dev/blob/aec3a901e6f6b3041b...
The codebase is a mish-mash of very old-style React+Redux code, with newer files and usages.
(I know this because I work at Replay.io, which started as a fork of the FF DevTools source, and we spent all of last year modernizing our fork :) Migrated the entire codebase to TS, ripped out all remaining uses of the ancient `React.DOM` helper functions and converted those to JSX, modernized the 2015-era Redux logic, etc: https://github.com/replayio/devtools/pulls?q=is%3Apr+sort%3A... )
In this case, I think it's also clear that most things you'd call an "application" abhor this behavior, which is why heavy clients are so popular, and small shops without frontend engineering talent are making a big push away from heavy clients which raises all the hubbub about HTMX etc
No they literally meant generating html like h1() rather than <h1></h1> in something like jsx templates
(I initially wasn't sure if they were talking about the model of server side rendering with client side events passed back to the server through a websocket either though)
lol, no he doesn't