>tried to adopt it, but it was (a) undocumented with no simple examples, (b) ever-changing, and (c) required you to use the Mozilla build system, even IIRC requiring you to basically build your code as part of the browser tree (I don't recall the exact details on this last point, but it was extremely difficult to use).

you're right, and I almost mentioned some of this. I developed some extensions in 2007-11, and lack of documentation was definitely a big issue.

I don't get it? I was excited about XUL back in 2007 (call me silly) and there was a whole resource site dedicated to it, complete with documentation and examples. XUL Plannet (https://web.archive.org/web/20070607174323/http://www.xulpla...) if I recall it correctly.

Back then, I wrote a pretty slick client for some RESTful API (I think it was a photo browser for flickr).

re: "undocumented". I bought whatever big oreilly book there was on it - "building XUL apps" or similar (https://www.oreilly.com/library/view/creating-applications-w...) it was impossible to follow. Granted, some new-to-me concepts and such in there, but after having respected colleagues try to work through the book with me, and they also gave up, I had to conclude this was not something worth pursuing. I mean no explicit disrespect to the authors; any book is a big effort, and the publisher sometimes overrides decisions or releases something too early. In any event, I vacillate between 'what a big missed opportunity' and 'we all dodged a massive bullet there' - I still can't decide which.

I also had that book. I think a problem with it was that XUL was moving quite quickly and the book itself was out of date.

I tried too and had the same experience.

XUL always felt like it was just made for Mozilla. It wasn’t but when you don’t provide documentation, stable APIs, and good tooling, that’s the message that you give.

As a user, you also know other users will have trouble too and that just tells you that it won’t take off.