VFS Implementation and user authentication

[Andrew Tridgell]

> Tim - this is why I don't want to ship vfs in 2.2.x, as
> we still need more implementation experience with it before
> we can freeze a first version of a vfs interface.

I think this is a reason why we should ship it! We just don't claim
that this will be the final version of the API. I don't think we
should ever "freeze" the interface really hard - I don't want it to
become our responsibility to remain compatible with every module that
people writes. We won't go out of our way to break it but we don't
want to get stuck with bad designs.

anyway, I think the VFS code has had enough testing for us to be
pretty confident that it doesn't break any existing functionality, so
I think that getting it in wider use would be a good idea.

I'm not the release guy for 2.2 though, so I disclaim all
responsibility if the world comes to an end when it is released :)