On Fri, Oct 31, 2003 at 10:27:22AM -0000, Richard Bang wrote:
> > > MODIFY id {size} data
> > Like Arnt said, this won't happen. I don't think it's much of
> > a problem just
> > to do this with APPEND + EXPUNGE.
> >
> Imagine I have an organisation with a 1000 people and each of them has a
> shared address book. Some of the entries are themselves shared as they
> relate to physical objects (maybe ldap data).Why don't they use one single shared address book? > In order to do append/expunge I would have to maintain a uid table for > each element for each user. This is silly. > I do not understand why an optional modify would not be permitted. So if > an entry is shared among 1000 users (in 1000 address books) I have to > create 1000 new UID's to update the object in all the mailboxes. I don't understand this. Only difference between MODIFY vs. APPEND+EXPUNGE is that the later changes the message's UID, nothing else. Why would a single contact be in 1000 different address books? How would MODIFY command know how to modify them all at once? I guess you're thinking something very different than what I am. You do mean that one "address book" mailbox would have one contact per message, right? > I'm not talking here about clients just dumping data into a mailbox and > saying "I'll remember that that's where I store addresses". But a > mechanism for a server to say to a client, "These are your address books > and none of the entries have been modified". > > Doing this with UID's means keeping a UID store for each address book > that each user can access and is a kludge to avoid having a modified > date. Only thing it'd require is for client to remember last known UID for each address book. If there were any changes, they'd have newer UIDs.
