Mozilla Developer Experimenting With Firefox UI In HTML

[smitty3268]

XUL currently is rather disliked by Mozilla devs for reasons I do not know(but I am curious what they are), so Servo does not implement and can not render XUL.

Although I'm not entirely familiar with all the details, my understanding is that it largely boils down to a lot of the same reasons X developers don't like X.

* Large amounts of legacy code, that's very complicated and no one really understands anymore
* Lots of strange performance cliffs you can run into out of nowhere
* Not efficient in general, and difficult to make so in spite of people putting in a bunch of work over the years
* Entirely their own custom code, which means no one goes to Mozilla already familiar with how it works, and no one goes there wanting to learn either - presumably most people working there are much more familiar and eager to work with HTML.

[smitty3268]

Years of work for the UI of a browser? Seriously guys (devs), unless you work for the government get your shit together.

I think you're missing the point, which is that html isn't currently good enough to do everything they'd want.

Presumably they would need to add a bunch of new features, and adding anything into HTML is a long drawn out process. Especially if you'd like to make it part of the standard that other browsers would implement as well.

[Marc Driftmeyer]

That's fine, as long as something replaces GTK2. Preferably something that actually plays well with configuration, like dark themes.

At the rate that WebKit2 GTK+ is headed I don't know why you would even give a rat's butt about Firefox.

I'm looking forward to dumping Firefox and using WebKit2 2.8 ready GNOME Web when it's fully baked. WebKit2 2.7.2 is out.

[liam]

Yes, and if the Servo browser is ready for prime time just one month earlier because they now spent some time to implement Firefox's UI in HTML, I'd say this project has been worthwhile.

I don't think servo is intended to replace gecko. We don't yet know Wyatt will work abs won't work. Rust itself is a work in progress with GC having changed a great deal.

[liam]

Firefox doesn't use the toolkits, so it won't work with them. This is just the idea of replacing the existing XUL renderer/interface with a HTML/CSS/Javascript renderer and interface. Mozilla's Firefox OS has an interesting aim - converge desktop and mobile apps to a common base of HTML/CSS/Javascript - and this is an extension of that. Everything in Firefox OS is HTML/CSS/Javascript, the launcher, the phone app, the camera app, etc. The hope is that standardised HTML/CSS/Javascript apps will replace a large amount of the existing Android/iOS/etc. apps and ease the pain of cross platform development. Native functionality (camera, GSM voice calls, etc.) is exposed through Javascript APIs and apps can run in a desktop web browser or embedded renderer.

The big question is whether the rendering and UI along with native JS (or cross-compiled asm.js style code) will be fast and responsive enough to replace native apps, and whether mobile developers will care enough to switch from native development. Google is pushing Java for application logic (with cross-compiling to Javascript and Objective C), and native UIs rather than HTML. Ximian is doing something similar with C#. Intel has its XDK cross platform HTML5 app runtime and tools. Many people have this idea of unified cross platform HTML/CSS/Javascript nirvana, though there aren't many Linux apps yet, or major apps on any platform for that matter - the most notable might be Youtube for the Playstation, which was written in AngularJS.

Rendering issues will be handled with servo or ganesh. Whether or not that turns out to be performant enough we don't yet know. My guess is that it will be, but it will also require help from web devs (upcoming best practices).
As far s applications are concerned I think we're going to see further movement towards relatively thick clients. Not over night, but within a number of years.

[Delgarde]

I don't think servo is intended to replace gecko.

Yes and no. My impression is that there's no actual plan at Mozilla for servo to replace gecko, no coordinated project to make it happen... for now, servo is more of an experiment. However, I do get the feeling that there's interest in seeing it happen at some point in the future.

[Delgarde]

I think you're missing the point, which is that html isn't currently good enough to do everything they'd want.

Presumably they would need to add a bunch of new features, and adding anything into HTML is a long drawn out process. Especially if you'd like to make it part of the standard that other browsers would implement as well.

Actually, I think it's almost the reverse - the feeling seems to be that at this point, HTML can already do most things better than the out-of-favour XUL. The problem is in first proving it (hence this project), and second, working out how to do a slow migration. Because it's not just enough to rewrite the Firefox UI to avoid XUL - that's relatively easy. The tough part is that they need to do it in such a way that all of the millions of Firefox add-ons don't need to be completely rewritten in one big bang cut-over.

[zanny]

At the rate that WebKit2 GTK+ is headed I don't know why you would even give a rat's butt about Firefox.

I'm looking forward to dumping Firefox and using WebKit2 2.8 ready GNOME Web when it's fully baked. WebKit2 2.7.2 is out.

These kinds of browsers already exist in the Qt world, like Rekonq or Qupzilla. The problem is that they don't have the addon infrastructure Firefox has, and a lot of the new top heavy web features like webrtc are complex monstrosities these browsers don't implement either.

[hoohoo]

I'm rather certain this is trolling, but I'm game, lets debate this.
In this case:
- what is/are the removed feature/s that set it apart from Internet Explorer?
- what is/are the added feature/s that make it more like Internet Explorer?

Apart:
- Close tabs button at end of tab bar. IE is per tab.
- Hide tab bar when only one tab.
- No status bar.

Like:
- paste and search
- paste and go
- removal of the things that set it apart

I don't really give a damn about paste-and-* except that in Mozilla forums their devs and PM-types claim these changes were to streamline the UI. I can paste a string then click Go or click Search - Go and Search have not been removed. Menu entries for same clutters the UI IMHO and strongly suggests some Moz devs playing dogma/power games (I know that is a strong statement, all I can offer to back it up is to go read the Moz dev forums. Moz has several devs who come on pretty sociopathic when defending these changes).

Prior to approx FF 29 I could change the behaviour of the tab bar in about:config. Now I cannot. Moz says to use one of several plugins or extensions - but all of them are buggy or flakey to varying extents.

Same for the status bar except it was removed circa FF 24 I think, and the extensions that Moz says replace it are all basically crap.

Status bar was extremely valuable place to present info about the STATUS of the browser.

I looked thru the code behind about:config once (some time ago). I don't think these things *had* to be moved out to extension-land. I think these UI changes were political/sophistic.

[hoohoo]

I think you're missing the point, which is that html isn't currently good enough to do everything they'd want.

Presumably they would need to add a bunch of new features, and adding anything into HTML is a long drawn out process. Especially if you'd like to make it part of the standard that other browsers would implement as well.

Thus XUL, no?