While I'm on the subject of Apple's doomed proprietary mix of C and smalltalk:

"In Objective-C 2.0, did Apple bother to elicit feedback from the language community? Or is it all Apple?"

Way back, in 1998, this guy, David Stes, wrote a very compelling paper about updating Objective-C to include smalltalk blocks. This would be a far more useful language feature than most of the garbage that they keep heaping into gnuobj-c.

GCC already has peculiar local function definitions (functions defined within functions). I'm sure since the compiler backend is already there, writing the parse tree should be a "no-brainer" for the rocket scientists who gave the Apple developer community a new Objective-C compiler when what they really need is a commercial-grade C++ compiler. Even Intel, a company which designs and fabricates silicon, is smart enough to own a state of the art commercial C++ compiler. Which begs the question: if GCC is so great and standards compliant, why would Intel bother releasing ICC for Macintosh? Or IBM, for that matter? Borland might make a good acquisition. Their C++ compiler technology is strong even if Microsoft showed them the door in WIN32. There's probably a reason that hardcore gamers (performance heads), continue to keep the Open Watcom dream alive. It's not like BSD doesn't have GCC available.

Speaking of Cocoa and Objective-C, I've updated cocoa.0x00000000.net with a sample that illustrates encapsulating new instance variables (private data) with informal protocols and categories. Source is included.