[citation][nom]mitch074[/nom]@TA152h: UNIX is a failure ever since it was created in 1970... Right.A little history.MSDOS was made as a dumbed down CPM - itself, a dumbed down UNIX. Remember, MSDOS was the OS MS sold for IBM, then added a GUI on top of it - up till Windows Millenium.When that dumbed down OS stopped being manageable, MS switched to their VMS-like kernel: NT. That one uses protected memory space, preemptive multitasking, a hardware abstraction layer etc. Like UNIX (as a matter of fact, NT can run a POSIX system: even NT ACLs have an eerie similarity to the draft POSIX ACLs).That so-called UNIX vendors failed to market it is one thing; that it failed is another. For one thing, most websites run on a UNIX or another. For another, OS X is a UNIX-like OS with a graphics layer that is not based on X11 - because remember, UNIX is not a graphical OS.But here, this is hardly relevant: this coalition intends to make a better experience for GNU/Linux (yes, I do mean the whole OS, not the kernel alone) on ARM systems. The problems: - the kernel isn't yet very optimized for SoC of the ARM type: eventhough Torvalds worked for Transmeta at one time, that part of the kernel still got less love than the x86, and would benefit a little from further development and testing. - the rest of the OS (starting with the libc library) is little or not at all optimized to make use of advanced ARM instructions: glibc, for instance, sucks on ARM (forks were made - it would be nice to have one and be done) - X11 has gotten modular, and can thus be kept, however, the multiple input driver is still stuck at the design stage, and it would be nice to have a unified one - due to X11's lack of MI, toolkits (gtk+, Qt, etc.) don't really have much developed to make use of these technologiesSo it's great to see big players working to make GNU/Linux more portable-friendly (and not merely more portable).[/citation]
Look at the way they are doing it.
Again a seperate place with seperate downloads.
I do find it's a noble and good initiative. Just that it isn't executed on a good way, it could be done much better.
Why don't they ask the Linux foundation for an extra clearly seperate branch of operation and fund it to do those things. This allows them to let things spill over to mainline more quickly and keep their own priorities. The problems you mentioned are not being solved. What they do is packaging and testing. This would be more efficiently done in strong cooperation with the Linux Foundation.
They should invest (actually starting with) more in optimization for and addition of low-level libraries for ARM at this point. Then working from low to higher in the stack until they reach the interface.
What Linux currently lacks is features and stabilization for low-level and high-level platform API's.