Hacker News Reader

Comment by lallysingh | Hacker News Reader

The first I see is the irony vis-a-vis pre-firefox Mozilla. They had a similar system (I'm looking for the old book I had on it) using a widget set called XUL. I tried using it. It hurt.

I guess the lesson is one of:

  (1) don't make complicated XML-based APIs
  (2) stick to standards (e.g., using HTML now instead of XUL)
  (3) if your system isn't getting traction, retry in 10 years?

View on HN

XUL is/was the language for the entire Mozilla app UI. So it's not like they invented it specifically for third parties to use- or even with them in mind.

XUL is only used for the external chrome interface of the Firefox desktop browser, let's be clear not to commingle the technology used to build the UI of Firefox with the standards-based technologies that power web apps/pages ;)

No, but it is entirely possible to build desktop-installable apps using the Firefox internals, with XUL as the primary interface technology. This was the idea behind XULRunner, a Firefox runtime package for you to build your apps with.

https://developer.mozilla.org/en-US/docs/XULRunner

The XML-based XUL isn't terrible, either, considering the respectable set of widgets and the ability to control all the styles with CSS. The nasty part, in my experience, was trying to get through the messy FFI that was XPCOM.

Check out Pencil (mockup app) http://pencil.evolus.vn/ for an example of multi-platform desktop app based on XULRunner.

I think XUL was/is vastly better then than HTML5 is today for what everyone seems to want: Actual applications (instead of web sites) that run in the browser.

XUL had/has a proper GUI toolkit with proper bindings to an event framework for user interaction. It was similar to HTML/CSS/JS but much, much better. The problem was people wanted "real" apps without having to do real work, and we're stuck with HTML which is a document language and will never be what we really want.

Oh, and it wasn't that complicated, it was rather easy.

I suspect that HTML was not mature enough back then which is why XUL was created and used instead.

In my opinion, HTML+CSS still remain quite inappropriate languages for describing desktop UIs. They can be used for such purposes, but, as compared with specialized languages (Glade XML, XAML, XUL), HTML+CSS is overly verbose for any non-trivial UIs, The worst thing is not code bloat and need for a Turing-complete language (JS) just to describe the looks, but the fact that you just can't have consistent system-wide look-and-feel. I believe, there's still completely no way to, say, bring proper system tabs or tree views to HTML+CSS+JS application.

Legit question: what about a tree view do you not think is possible with HTML/CSS/JS?

Native look and feel.

It's interesting how the traditional definition of 'native look and feel' is starting to seem passé. Most of the computers I have sitting around the house today do not utilize a tree view in any default 'app'. There are no interface components on my tablet (that I can think of) that can't be recreated with some combination of HTML, CSS and JavaScript in a modern browser.

That's fair, but other than that, I'm curious what there is. CSS is still bad at grids without pretending they're tables, but Flexbox has solved most of the other limitations.

Keyboard acceleration. I have never seen an HTML control kit that handles the Tab or Arrow keys correctly or even very well.

No standards. For example - Drag and Drop. There is no standard system for Drag and Drop between components in the browser because it doesn't know about these controls, it only knows div div div div div. So, in general - it's just harder to write good code that targets HTML components because there are no standards for complex components.

(Some more patterns with no standards for complex components: Data Binding, Encapsulation (grouping one or more component into a parent component and walking parent/child components), Inheritance.)

And then, there's performance. Native trumps HTML here, no contest.

When you say:

"(2) stick to standards (e.g., using HTML now instead of XUL)"

and

"(3) if your system isn't getting traction, retry in 10 years?"

It sounds like you are claiming Mozilla is promoting the use of XUL as a mark-up language for web apps, which is completely, unequivocally, 100% false. Hopefully this is not what you intended, because it would be an unfortunate case of FUD-spreading.

It's not. AFAICT, they were promoting XUL as a way to build desktop apps. I even have a copy of the O'Reilly book about XUL... somewhere.

(3) was talking about making apps with mozilla. Not specifically xul.

Reply on news.ycombinator.com