>On Mon, Jun 04, 2007 at 10:39:16PM -0700, Scott Rotondo wrote: >> There are certainly other use cases besides CIFS for "system attributes" >> on files. Does this case establish a protected namespace (e.g. SUNW*) >> for attributes whose meaning and access control are determined by the >> system? If not, could the case be expanded to do so? > >Rich explained to me that the intent of these SUNW* attrs is to make it >possible for existing archivers to backup these new attributes without >having to know about the new APIs and without having to represent >things like IMMUTABLE as Solaris extended / NFSv4 named attrs.
How does the project team intend to solve the fact that such attributes have no meaning to other implementations and that restrictions which need to be applied on modification of such attributes in the context of Solaris are also enforced in other environments? Suppose I have a filer of type X; I've added a SUNW_IMMUTABLE attribute from Solaris to a file from a Solaris client. What is stopping a Windows client from removing that attribute? Casper
