I see Rajiv's proposal as an instruction to the file system on whether
metadata on the file's organization and contents is to maintained in
persistent storage. (Correct me if I am wrong, Rajiv.) Clearly such
information is needed but I don't understand why this information
should be maintained by the file-system. I would argue that such
information should maintained by individual applications and/or
libraries. The list mentioned above goes far beyond the original
proposal. These purposes can be easily served by application-specific
conversion programs that can read the metadata (eg HDF).
anurag