Shoehorn := Longhorn - WinFS
The intention is good. The choice of delivery vehicle (i.e. a "file system") is not so good. Once again, the problem seems to call for a simplification of dynamic, semi-structured, shared databases rather than a shoehorning of these concepts into a future Longhorn file system.
The advantages of the database approach has engineering merit as well as marketing.
Perhaps there was marketing merit in moving O/R mapping and business domain frameworks into WinFS. I didn't see it, and I certainly don't see the engineering merit.
Will we see ObjectSpaces and MBF pulled back out into cross-MS-platform capabilities? Avalon has joined Indigo in this regard. Why not the rest?
MSFT should make WinFS a across-MS-platform-or-wider improvement on *databases* rather than file systems, and all the Longhorn components make engineering sense, probably marketing sense.
And MSFT should make this an opportunity (along with IronPython) to continue down the path of a post-modern approach to systems development.