Now, it may be that in the interest of expediency and quick ROI, they may
just slap NS on boxes and leave it at that -- full blown personality
change. Some NS'ers may even want that. I think it may be more dramatic and
certainly more artful *not* to revolutionize the UI. What NS offers is
*solidity* : in the OS, in the responsiveness of apps, predictability of
performance, even the graphical solidity of window movement. How much more
pleasant the fit and finish of MacUI would be if indeed, the illusions of
movement, of switching, of multithreaded behavior were reliable enough to
become real? (e.g. even something as simple as a background thread
reporting current disk capacity in the Finder frame -- something Win95
fumbles too.)
What makes an interface more "real"?
Returning to Simson's pieces: filters, faxing, i18n, NSBundles instead of
resource forks are all major plumbing wins and will surface in the UI. I'm
doubtful of the dock, though... but that may be unavoidable. (Remember, the
dock is part of the 'raising the lowest common denominator heritage': you
can afford it on a megapixel display and *need* it to manage >50 windows
(my avg, it seems) -- it's not suitable for 640x480
Now, how could this be implemented? Well, the appkit was already rewritten
once to be more UI-netural (OpenStep), and even reimplmented with
partially-native widget sets (OpenStep/NT). With some extra resources in
AppKit engineering, it could be done for MacUI, too. The only other hole is
rethinking some UI idioms of NS that have been coopted over the years: e.g.
Inspector-style nibs, which cannot be easily be converted to tabbed-dialogs
because Inspectors are a hack, not an Appkit feature. Sure, modern OS has
tab idiom all over the place, but how many apps have been written from
scratch for 4.0?
Of course, some things like NXBrowser should be promoted. 'Shelves'
probably shouldn't.
RK