Last week, via the official MSDN WinFS blog rather than the usual Microsoft PR channels, it was announced that WinFS is effectively dead. There was much talk of the great many technical innovations the WinFS project has created and how those innovations may find their way into a broader Data Platform Vision and examples cited such as the Entities features being built into ActiveX Data Objects starting out as something for the WinFS API. But at the end of that much talk the bombshell remained: "we are not pursuing a separate delivery of WinFS, including the previously planned Beta 2 release.” In effect WinFS has not only been downgraded from optional download sometime after Vista hits production, to a possible component within future editions of ADO.NET and SQL Server, but made extinct for all real world intents and purposes.
So much for replacing the legacy Windows file-system, one of the oldest parts of the platform, with a search ready, object-oriented engine that could be queried with a SQL variant (think SQL schema instead of File Allocation Table) and truly revolutionize file storage and retrieval for Windows users. So much for the much heralded three pillars of Vista which is now down to just two: Windows Presentation Foundation (Avalon) user interface layer and Windows Communication Foundation (Indigo) web services layer. So much for the object-oriented file-system first promised, and tantalizingly teased in preview releases, back in 1993 when Cairo was still an ongoing project.
By way of a response to the somewhat angry and confused reaction from developers and users alike, the official MSDN WinFS blog issued an update to the update this week. Unfortunately it just manages to muddy the waters even further, or as one developer described it "putting PR spin on the PR spin." Quentin Clark, who runs the WinFS Program Management Team from Building 35 on the Microsoft Campus, is responsible for directing the team that has been building WinFS. Or at least he was, his future must surely now be in doubt. Although that rather depends upon how you interpret such statements as “is WinFS dead? Yes and No. Yes, we are not going to ship WinFS as a separate, monolithic software component. But the answer is also No - the vision remains alive and we are moving the technology forward. A lot of the technology really was database stuff – and we’re putting that into SQL and ADO. But some of the technologies, especially the end user value points, are not ready, and we’re going to continue to work on that in incubation. Some or all of these technologies may be used by other Microsoft products going forward.” So that’s cleared that up then…
By claiming that the dropping of WinFS has no impact upon Windows Vista because “we announced back in August 2004 that WinFS would not be in Windows Vista” and avoiding the question of whether a relational file-system will ever be in Windows now with “hey – we are very busy finishing Vista, and just aren’t ready to talk about what comes next” Clark has done neither himself, nor Microsoft, any favors. The ‘no impact on Vista’ statement is, frankly, a crock. Of course it impacts Vista, WinFS has always been seen as integral for the future OS vision that Vista is meant to introduce. Plans were always, as far as most observers, developers and users were concerned, in place to release WinFS for the Vista platform. Indeed, Microsoft themselves were still talking it up as recently as at TechEd 2006 a couple of weeks ago!
There simply appears to be no commitment to a relational file-system for Windows now, and the fact that some of the WinFS technology is likely to find its way into SQL is small comfort for developers and none at all for the end user. I’m left with the inevitable conclusion that Microsoft simply finds it too hard to upgrade the file-system, and is equally inevitably losing ground to Apple in this regard. What happened to real innovation from Microsoft? Not just the Windows dressing that is Vista, but real core platform innovation?