David Powell wrote:
If we are going to support the option of the client supplying an id,
then we can't support using a bogus id, else the how is the server
supposed to tell whether the id is intended to be genuine or not?
If we want to allow entries to be reposted in different collections or
servers, then we must allow the client to supply the id, and the
server must not be allowed to silently overwrite the id with one of
its own, else subscribers will see duplicate entries etc...
http://www.imc.org/atom-protocol/mail-archive/msg02013.html
“Maybe a special value of atom:id in the initial POST could trigger a
"generate a cool atom:id" on the server.”
Answer from Joe Gregorio:
http://www.imc.org/atom-protocol/mail-archive/msg02014.html
“That's an interesting idea”
--
Thomas Broyer
