+1. This whole discussion is a rathole. Sometimes the client
should|can|will generate the id, other times the server should|can|will
generate the id. We should not dictate that the client must be
responsible for generating a globally unique identifier for every
atom:entry it wants to post. But we should also say that that
collection must contain valid atom:entries. This means that for POST'ed
entries that do not contain atom:id and atom:updated, the server MUST
fill them in upon during processing of the POST operation.
Joe Gregorio wrote:
On 11/1/05, Luke Arno <[EMAIL PROTECTED]> wrote:
What about "moving" entries from collection to collection?
What if I copy an entry from one collection to another
in order to begin editing a whole new entry, that is, I use the
original entry as a template or boilerplate? How is the server
to distinguish between the 'intent' of the two kinds of copies?
If it were just a copy then you'd want the
server to keep the atom:id it was supplied with, in the
case of boilerplate you'd like a new atom:id supplied.
Trying to legislate anything above "atom:id is server controlled"
will turn into a rathole.
-joe
--
Joe Gregorio http://bitworking.org
