Future OS X system API?

Okay, so a chap named John Siracusa wrote this for Ars Technica:

http://arstechnica.com/staff/fatbits.ars/2005/9/27/1372

The gist is that he believes the Achilles heel of OS X is that in these days of CLR/MSIL/C# and JVM/bytecode/Java, we're still using Objective-C which is too low-level for use in a modern OS API (lacks memory management, a VM, etc).

As a guy who once did a little research in the areas of concurrent and distributed programming, I'd love to see a virtual machine that actually guaranteed safe program operation in a symmetric multi-processing environment. By which I mean that the program works "as expected" regardless of what type of OS X beast you run it on -- precludes race conditions etc. The guarantees made by VM specs such as JVM and CLR, btw, are not very strong when you get right down to it. And many implementations of these specs (for Java at least) don't even meet these weak guarantees.

I think Apple could get a lot of mileage out of embracing a combination of a strongly typed object-oriented programming language and a dynamically typed object-oriented scripting language in a new OS X API. Like Apple's products, these languages should be elegant and clean, achieving what other languages do but with more style so that the software engineering task is more enjoyable on a Mac. I think Eiffel and Ruby would be a good combo to consider. Software written for the new OS X API would end up being more stable, reliable and robust than that written for other platforms.

Here's the tricky bit though -- Apple should port the API to other platforms so that software written to target their API can (more-or-less) run on, say, the CLR. All of a sudden, developers can create software that interacts with a nice clean system API and yet doesn't need much/any porting to target both Windows and Mac. Oh yeah, and it just so happens to run natively on the Macs. Most system APIs are clunky, so if Apple got this right then a lot of developers might give it mindshare.

Although a lot of Mac software developers sound quite content with Objective-C, but I think I see where John Syrcusa was going with this. I would estimate that the flow-on benefits for end-users of having a bunch of programmers who are really gunning to use your system API rather than anyone else's, could be really substantial. Sort of like the benefit of being the first guy to make 1.8" hard drives -- sure resulted in a lot of happy iPod customers...