Re: Updated metadata proposal

[Tom Tromey <tromey cygnus com>]

>>>>> "Felix" == Felix Bellaby <felix@pooh.u-net.com> writes:

Felix> The idea of using garbage collection is attractive in terms of
Felix> its ease of use but I wonder whether this is a sensible
Felix> approach to take.  Garbage collecting metadata on a multiuser
Felix> machine with large file systems and extensive VFS references
Felix> could take a long time and is particularly unattractive for
Felix> files held at remote ftp sites.

I agree that for remote files this is a problem.  However, some kind
of GC for the metadata is required, because metadata is held per-user.
If you assign some metadata to a system file which the sysadmin
deletes, then what?

Felix> We are already relying on applications to record file creation
Felix> and movement. At the very least, the API should include a
Felix> command to clear out all the metadata for a file when it is
Felix> removed.

The API has this.  But you can't clear out anybody else's metadata.

Felix> If garbage collecting is used then a command to invoke it on a
Felix> directory tree might be useful in the API so applications can
Felix> help suggest when it would be useful.

I agree.

Suggestions for this are welcome.  I'm not sure the best way to go.

Felix> I am not sure about distributing the metadata db files over the
Felix> user's filespace.

Me neither.  The first implementation won't do this.

Felix> Multiuser systems could benefit from having system
Felix> "metadatabases" (in /etc and /usr/share ?) which would provide
Felix> metadata defaults for shared files.

Yes, I agree.  I've added this to my to-do list.

Tom