Magenta Pairs Linux With Darwin/BSD, Is Like iOS

[uid313]

I've wondered about this as well.
Apple has really great docs, possibly as good as msft.
Add to this someone porting iokit to linux and then you have something of real interest (though I wonder if reimplementing cocoa would be legal).

Didn't the court rule in the Oracle vs. Google Java/Android case that APIs are not subject to copyright?

[FireBurn]

Didn't the court rule in the Oracle vs. Google Java/Android case that APIs are not subject to copyright?

No - it didn't go quite as far as that. It was to hopefully prevent the ruling being overturned. The ruling was case specific

[M1kkko]

What would be much more interesting, would be an environment where you could run OSX applications on top of a Linux kernel, maybe chrooted beside a simultaneous GNU userland.

That would be nice, even if it was just a compatibility layer like wine. What surprises me is how nobody has really attempted to do this, even if it were for free-BSD

Doesn't such a thing exist already?

Just after a bit of research: http://gitorious.org/ringo

"Ringo is a runtime environment that runs Mac Applications on GNU/Linux"

I haven't tried that myself and I don't know if it works yet. But at least it's work-in-progress.

[ninez]

Doesn't such a thing exist already?

Just after a bit of research: http://gitorious.org/ringo

"Ringo is a runtime environment that runs Mac Applications on GNU/Linux"

I haven't tried that myself and I don't know if it works yet. But at least it's work-in-progress.

Nice find.

Ringo actually looks much more interesting than Magenta, since it aims at running MacOSX applications on Linux. (x86/x86_46)

it says in the README a few things are working. You able to compile things like libmach.so, libCoreFoundation.so, etc.... I wonder how much work it would take to get it to a point, where you could run a native MacOSX application? (even one that doesn't heavily rely on Mac-specific APIs?).

it also doesn't look like there have been any commits in about a month.

I bookmarked the project anyway, just to keep an eye on it.

cheerz

[staalmannen]

That would be nice, even if it was just a compatibility layer like wine. What surprises me is how nobody has really attempted to do this, even if it were for free-BSD

If I understand it correctly one of the most important component for such a layer is provided in this project : mach compatibility and probably the dyld file format.

[droidhacker]

Doesn't such a thing exist already?

Just after a bit of research: http://gitorious.org/ringo

"Ringo is a runtime environment that runs Mac Applications on GNU/Linux"

I haven't tried that myself and I don't know if it works yet. But at least it's work-in-progress.

That's definitely not what I have in mind. That's like Wine.
There are three possibilities for running OSX applications on Linux;
1) Wine-like emulator (bad for the same reasons as Wine is bad... nothing works.)
2) Virtual machine (way too much overhead, zero integration...)
3) Take existing workable OSX userland and run it on a Linux kernel.
#3 is what I'm talking about. A similar thing is done for running debian/GNU userland on Android devices: http://www.mayrhofer.eu.org/debian-on-android
... of course the advantage there, is that there is no problem interfacing that userland with the Linux kernel.

Basically, we don't want to have to run a second kernel, and we don't want to be stuck with the hackish wine-type crap where everything has to be reimplemented the hard way.

[Teho]

Basically, we don't want to have to run a second kernel, and we don't want to be stuck with the hackish wine-type crap where everything has to be reimplemented the hard way.

How do you imagine doing that? Most of the libaries that application actually interact with in OS X are propietary.

[ninez]

That's definitely not what I have in mind. That's like Wine.
There are three possibilities for running OSX applications on Linux;
1) Wine-like emulator (bad for the same reasons as Wine is bad... nothing works.)

This is almost certainly what i would like, minus nothing working (of course). For me, Wine actually works very well. Although, most of my usage is proaudio plugins, which many at this point (both free and commercial) work very well. I know a few other people who successfully are running (other types of) windows applications in Wine that work, and a reliable. but obviously there are many that still do not work, or are buggy.

2) Virtual machine (way too much overhead, zero integration...)

MacOSX virtualized also comes with some limitations (as all VMs do), the main one being no Quartz-extreme and thus any apps that require it will not work, with a few exceptions that have been made to work. You can get (some) 3d acceleration, using vmware svga II drivers for mac;

http://sourceforge.net/projects/vmsvga2/

You can get some integration using the Darwin vmware tools from VMware fusion for Mac, ie: drag drop, auto-resize, etc. while also being able to use MacOSX (native) drivers for attached devices/hardware. But obviously, overhead is always an issue - although a Win7 VM sucks up way more resources than MacOSX - and performance (aside from some animations, due to lacking QE) is actually decent.

3) Take existing workable OSX userland and run it on a Linux kernel.
#3 is what I'm talking about. A similar thing is done for running debian/GNU userland on Android devices: http://www.mayrhofer.eu.org/debian-on-android
... of course the advantage there, is that there is no problem interfacing that userland with the Linux kernel.

Basically, we don't want to have to run a second kernel, and we don't want to be stuck with the hackish wine-type crap where everything has to be reimplemented the hard way.

I actually tend to think #3 is the hardest way, and with lots of changes probably required to the linux kernel, itself. A lot of kernel interfaces would probably have to be modified in order to support various things from Darwin/MacOSX. It would probably require having a huge delta to maintain, as i can't see upstream linux development supporting it. as well as other technical (and probably legal) issues to boot....

1 and 2 also offer some advantages that 3 doesn't, as your third option would mean gnu/linux software would become a 2nd class citizen (or non-existent), ironically -> on a linux system you'd have macports (in your Mac-userspace) for some stuff, but i would prefer to keep my Linux Desktop intact (otherwise, i might as well just use MacOSX and it's proper kernel). Access to gnu/gpl/free software is much better on most (good) distros over MacPorts, as well.

I actually hope this Ringo Project turns into something, that i might actually be able to make use of.

Where as your #3 seems pretty much pointless, and would yield no benefits. In the case of Debian on Android - it actually makes sense, and would be good - as lots of people would like to be able to run your standard gnu/linux desktop + applications on their Android device.

Can you give a good example, of how running MacOSX on linux would actually be useful?

[curaga]

Native access to Photoshop, et al?

[liam]

Native access to Photoshop, et al?

Come 2.10 (which should be coming rather sooner than the last release, unless something unforeseen happens), GIMP will have closed a significant gap with PS. What's more, the filters will start to look far better just by enabling 24bit color.
What I would like from Mac are some of their little goodies like TextMate.
Access to some of their media tools would be a bonus, though I do hope that major apps like Avid, Vegas, PT move to the new gstreamer sdk.