why not palringo?

Looks like there's already some support or stubs for support in place, from vanilla-latest Leopard (maybe I'm special because I have Developer tools):

/System/Library/Filesystems/zfs.fs
/System/Library/Extensions/zfs.readonly.kext
/usr/lib/libzfs.dylib
/usr/sbin/zfs
/usr/share/man/man8/zfs.8
/usr/share/man/man8/zfs.util.8

Probably just read_only support so we won't complain about getting at ZFS stuff from our old machines, etc.

I think the screenshot here sums up my issues with it. That was with a somewhat reduced catalog, somewhat reduced plugin set, fairly fresh after a quicksilver restart. Triggering it would cause a huge amount of thrashing, to the point where using spotlight is faster as an application launcher.

Maybe I'd install this if it worked with Facebook status too.....

No Exchange or Office support. -- Dude, we haven't heard anything about third party apps on this thing, but from the fact that it runs OS X, I would not be surprised to see 3rd party applications ported with just a replacement NIB and a recompile for whatever cpu this thing uses. We'll have to wait and hear what they're going to tell developers though.

My prediction for the end of the year is more two way interoperability between second life and other internet protocols and applications. The most obvious thing for linden lab to do is add Jabber messaging, but I'm sure second life objects will begin to integrate with external 3rd party applications more.

Is there really that much of an advantage to these builds? GCC has to know how to do special magic for these machines. Is there something being done that gives an overall benefit from a simple optimization technique between the G4 and G5 (for example). Perhaps the ability to use more registers all the time or addition of significant amounts of SIMD instructions?

Also, I understand how we can compare optimization of algorithms, but does anyone have a good metric for how optimzation benefits a whole application?