I'm -1,000,000 on using POST for this feature. Too bad, but since Atom is so open I'm free to try it without the approval of this WG.
Walter Underwood wrote:
I strongly favor POST over a new HTTP method. Post has existing semantics and parameter-passing formats.
POST doesn't have any semantics. It's application specific. The protocol even uses POST to DELETE in the SOAP part.
Relying on those makes our spec smaller and implementations easier.
New methods aren't hard to do. This is just FUD.
Defining those anew for Atom makes the spec bigger and implementations harder.
No, it makes this feature insecure and opaque.
POST should also make QA and load testing easier. Anybody got a load tester that already implements ERROR?
Writing a switch statement at the method level is harder than writing a switch statement to figure out what POST *really* means? Nah.
Who in the world would use a load tester that doesn't let you configure every aspect of the request. How else will you know how your server reacts to methods it doesn't support?
Robert Sayre
