As Tim Bray says “Wow”. Here is the announcement post with a huge number of comments. This is discouraging. As I have argued before, we need the kinds of capabilities WinFS was striving for to make the next leap in managing information. I remain skeptical that database platforms are a sufficient solution for effective object management – they may be the necessary next step, but they are certainly not the ultimate answer.
There are no doubt many easier, shorter-term ways to get return on software development than a radically different operating system, but hopefully at some point there will be sufficient recognition by all the software infrastructure vendors that working together to build a modern OS would be worth it. On the other hand, perhaps what has happened to WinFS is really a sign that the days of huge operating systems are numbered. The problems are really bigger than any one platform. What kind of cross-platform infrastructure is feasible to accomplish the fluid, granular and meaningful interchange of content and behavior we know we need? This is a more interesting question than whether WinFS itself is dead.
UPDATE: There is a lot of commentary out there, but as usual Jon Udell has a view worth reading.
It is good to see that MS is plugging away at this and now providing a way to really take a look at it. I am not qualified to comment on its technical merits, but the idea is important (to content management but more to computing in general) and something we have been watching for a long time. Dan Farber’s entry on this release has some useful links, and Wikipedia has an accessible intro to WinFS
In an earlier post on Longhorn adoption, I talked about the need for an operating system that provided support that went beyond simple file management to include services that content applications could leverage. Will Longhorn’s WinFS do this? Will other operating systems?
One of the questions we’ll be asking our panel on the Future of Content Management at our conference next week, will be “Where in the software stack is the best place to provide basic content management functionality, e.g., content elements with attributes and metadata?” With senior strategists from Oracle, Interwoven, FatWire and Mark Logic on the panel we ought to get some interesting discussion going. If you have a question you would like to see us address, comment on this post or send me an email.
In my next post I’ll look at how this question relates to one of the fundamental issues underlying the keynote debate on XML Strategy and Open Information.
Dan Farber raises the issue of Longhorn adoption and quotes a Jupiter analyst who claims the challenge is that XP is “good enough”. There is actually a more fundamental reason the question of adoption is interesting. What is that and what does it have to do with content technology?
I’ll start the answer with a little history. In 1994 at our first Documation conference, I moderated a debate between Tony Williams, Chief Architect of COM at Microsoft, and Larry Tesler, Chief Scientist at Apple. The Microsoft COM and OFS/Cairo and Apple OpenDoc efforts both recognized the need for operating systems to provide more support for the richness of unstructured information than is possible with the primitive file systems we had then.
Before the debate I preferred the OpenDoc approach because it seemed more consistent with my view that new operating systems needed to be able to manage arbitrary information objects and structures that could be described with a markup language (like SGML at the time). However, Tony convinced me that OpenDoc was too radical a change for both users and developers at the time. Tony agreed with the ultimate need to make such a radical change to file systems to support the growing need for applications to manage more complex content, but he said that Microsoft had decided the world was not ready for such a shock to the system yet, and defended their strategy as the more realistic.
Eleven years later and we are still stuck with the same old-fashioned file system in spite of the fact that every modern business application needs to understand and process multiple types of information inside files. This means that database platforms and applications need to do a lot more work than they should to work with content. I am no expert on Longhorn, but the file system that will be part of it (although maybe not initially), WinFS, is supposed to go a long way towards fixing this problem. Is the world ready for it yet? I hope so, but it will still be a big change, and Tony’s concerns of 1994 are still relevant.