On 10/27/05, Eric Scheid <[EMAIL PROTECTED]> wrote: > > On 28/10/05 11:42 AM, "Luke Arno" <[EMAIL PROTECTED]> wrote: > > > It occurs to me that in most mail clients, drafts are > > just kept in a folder. Why not just have draft > > collections? Have we talked about this before? > > we have discussed it before, but at least three problems arose > > 1) how to move an entry from the draft collection to the published > collection >
POST, DELETE? The question that leaves me with is: What about atom:id? Is it OK to think of the draft entry and the published entry as two independent entities? > 2) some implementations want more than draft vs published > That actually sounds like an argument for this approach: a collection for each state in your state machine. > 3) the idea of using collections for segregating according to status led to > the idea of collections segregating according to author, category, and other > meta data, which led to the problem of exploding combinations > But we already have that metadata in atom syntax for all those things so those would be less sane. There is nothing to stop implementations from doing this unless we make the APP very stiff, anyway. In fact, I think it is inevitable that the APP will be used for queueing up entries for multi-stage processing in this way. - Luke
