zipfs and relative paths

[Joel Uckelman]

Thus spake Alan Bateman:
> Joel Uckelman wrote:
> > The zipfs demo always returns absolute Path objects. Can anyone comment
> > on whether this is (a) a feature, or (b) that relative Paths are ok, just
> > not implemented yet?
> >   
> I remember Rajendra fixed a number of bugs in this area but they may be 
> more. In particular, resolving a relative path against another relative 
> path should yield a relative path. Also, invoking newDirectoryStream on 
> a relative path should yield entries that are relative paths. If you see 
> otherwise it's a bug.
> 
> -Alan.

What about paths ending with '/' (which are not just equal to '/')? Right
now, normalize() maps 'foo/' to itself. I would not expect an Path to
contain redundant separators after a call to normalize. (This particular
behavior is documented in the comment immediately preceding normalize().)

-- 
J.