I've been looking at this project for a couple of months now, mostly because it's the closest you can come to OS X and the Cocoa framework on your Linux PC.

A few details for those who'd like to know: GNUStep is a reimpementation of the OpenStep API, which was an effort by Apple and Sun to bring the insanely great NextStep API to your regular workstation (I think my facts may be a little wrong here ; feel free to correct me). Etoile is built using GNUStep, with the desktop model based around the OS X dekstop (check out my desktop and compare it to Etoile: http://img136.imageshack.us/my.php?image=picture1rn3.png).

The best part about Etoile is not that you get to use a fast, lightweight, (sorta) modern desktop environment for your favorite open source OS, it's the fact that you get a great oject oriented API similar to Apple's Cocoa for writing applications with. I've been looking at Cocoa for the past month or so, and I can assure you it's a great framework to write desktop apps in. Sure, Objective-C can be a little weird at first, but once you get used to it you'll find it's a very well thought out language. And BTW, GNUStep even has replacements for Apple's XCode and Interface Builder. I don't quite remember what the XCode replacement was called, but the IB replacement is called Gorm (which is not a cool name).

Upvoted for awesomeness.

So there's no way you could just use Cocoa? Maybe it's not tinkery/Free enough? :)

I do use Cocoa on the Mac, but sometimes I go, "What if Apple decides to drop this stuff, like what they did with Carbon?", and I shudder.

Then your internalized knowledge is something worth for doing webstuff with SproutCore.

That certainly looks interesting. I skimmed through the Hello World tutorial, and it looks like they use many of the same concepts as Cocoa. Will keep this in mind when Apple drops Cocoa ;)

Didn't they drop Carbon to force people to "switch to Cocoa already!" ?

This really needs a video of some sort to show the desktop in action. From just reading things and looking at screenshots, it looks like it provides a better platform for developers but is a quantum leap backwards for users. I assume applications have to be specifically developed for the platform, so that cuts out 99% of the applications users are going to want to use (Firefox, Gimp, OOo, etc.).

Still, if it's really a nice platform, I hope it would become popular enough that popular applications could be easily coded to work with it.

It runs on top of X11 just like every other Unix GUI, so Firefox, Gimp, OOo, etc will all run just fine.

They'll just look different from the etoile applications.

Looks like the average linux desktop - in 2001.

I think it's only partially about the desktop and more about the experience. I'm fairly certain they're looking for help, you know, if you'd like to make it look like something in 2009, or maybe something for the future!

I don't believe in building the umpteenth desktop environment unless you can add something new to the pool. There are OSX style themes for all major window managers and many of them look way better than what I see in those screenshots.

I could turn your question around: Why did they start their own instead of just supporting, say, XFCE?

> Why did they start their own instead of just supporting, say, XFCE?

Because XFCE doesn't have the same goals as they do. If you listen to the interview with them on FLOSS Weekly at twit.tv, they talk about how they want this thing to grow, and what they want it to be. The best example I guess I could give is that they are creating a desktop environment based on concepts in Smalltalk. The look isn't the important thing (at least right now). The cohesiveness of it all is what's interesting.

They're building on top of GNUstep.

Because getting rid of X would be a win?

It doesn't get rid of X, it sits on top of it just like XFCE, KDE, Gnome, etc.

It's easy to say that getting rid of X would be a win, but replacing it at this point would be much more work than fixing what's wrong with it.

Any chance of a version (or even a separate project) aiming at binary compatibility with Mac OS X? I think a Wine-style project for Mac OS X would more a more interesting and popular development than a niche platform.

Cocotron -- http://cocotron.org -- is a project that goes a fair way in that direction, although through source compatibility rather than binary compability. It's two things:

  1) An MIT-licensed reimplementation of Cocoa (Foundation + AppKit + some CoreFoundation plain-C APIs)

  2) A cross-compiler GCC toolchain that produces native Win32 binaries and installs into Xcode. (There is also some degree of Linux support.)

The benefit of this combination is that you can basically add a Windows target to an existing Cocoa project in Xcode and have it produce an .exe file... At least if all the methods used by your application happen to be implemented in Cocotron.

There's still quite a bit of the Cocoa API missing, but the amount of work already done is impressive -- and one of the really great things about Cocotron is that it feels so easy to contribute to the base frameworks. The Foundation and AppKit projects are clean of cruft, neatly laid out, and they are built as Xcode projects so there's no toolchain learning curve: if your favorite esoteric NSString method is missing, you can simply open Foundation.xcodeproj and implement it.

For me, Cocotron is cross-platform Objective-C that "just works". YMMV.

In their recent interview at http://twit.tv/FLOSS, they mention that it might possibly work with OS X in the future if some issues with the linker on OS X are resolved.

Compared to GNUstep it looks nice, but that's not saying much. GNUstep is incredibly dated and downright hideous: http://www.gnu.org/software/gnustep/images/full-screenshot1....

There's a conversation with the developers on TWIT's FLOSS podcast: http://twit.tv/FLOSS

The thing that's sort most interesting about this is CoreObject, which is modeled off of the ideas of SmallTalk images.

This is really quite cool. I think it has potential, and if they can truly unify user experience application-to-application like OS X does, Linux has a chance at more desktop uptake with the non-hacker population.