Regardless of my personal opinion, for our implementation it really doesn't matter. The spec should just be tightened to make it a bit more explicit.
- James Thomas Broyer wrote: > 2006/4/27, James M Snell <[EMAIL PROTECTED]>: >> In our case, at least, the read-uri and the edit-uri are the same so >> it's not that big of an issue for us. Either way, however, the spec >> should be updated to help ensure that implementations behave consistently. > > +1 > >> If the entry in the response body is optional, the Location should >> likely always point to the edit uri. > > Why? Aren't you going to retrieve the entry anyway before editing? > > What if the entry is not editable? > >> If the entry in the response body is mandatory, it really doesn't matter >> what the the Location points to. > > IMO, whether or not the response body is mandatory, the Location > header MUST be an URI you can dereference and GET to retrieve the > entry. The "edit uri" would then be in that entry in a > [EMAIL PROTECTED]"edit"]/@href. > > With GData's optimistic concurrency mechanism, you'll have to GET the > entry (at the Location or [EMAIL PROTECTED]"edit"]/@href URI) before starting > editing to make sure you have the latest version and thus the latest > edit-uri. > (Note however, that you shouldn't GET the entry after editing but > before PUTting the edited entry or you might have lost updates) > > -- > Thomas Broyer > >
