I've been watching - and reading - the recent debate over the announcement to postpone the launch of Objectspaces to Longhorn by aligning it with WinFS.
Barry Gervin has a report from TechEd on this issue that ties together this decision with MBF also being moved to the Longhorn timeframe. I think it makes sense to make every effort possible to make sure that when we [MSFT] decide to put out a new data-access API, we can say that we have done everything possible to make sure that it is aligned across all layers that will build on it, and that it will provide a stable programming model for many years forward.
I don't think it is prudent to dismiss WinFS as "Just a Filesystem" that does not have anything to do with data-storage. It represents a paradigm shift in terms of connecting structured and unstructured storage, database and filesystem, and provides an object-relational model for data access - That, combined with the need to support the requirements for data access that ObjectSpaces and MBF bring to the marriage will help make WinFS suited to be the data-access platform for the future. For MBF this partnership will surely mean that we can now rely on a strong underlying persistance platform that will also be the data-storage platform of choice for Longhorn generation apps.