Tim Bray wrote:
I've never thought there was anything terribly wrong with WebDAV in theory. In practice I've had all sorts of pain getting clients and servers to interoperate, it sure ain't a "Just Works" kind of thing....
Yes, there are many bad client and server implementations out there. (yes, I'm looking at you, Microsoft and Apple :-).
AtomPub is different from DAV in two key respects: - The client doesn't control where things go, the server does
Yes.
- It is allowed and expected that an AtomPub server will look at the incoming information and change it (generate ID, timestamps, sanitize HTML, etc)
Yes and no. It's true that there are clients that assume that anything that smells like WebDAV behaves like a filesystem, but the specification doesn't guarantee that, and there are lots of cases where WebDAV *is* used for other cases. One well-known example is CalDAV (in which the server generally *will* sanitize calendar data).
Both of these things probably make AtomPub a better match for the needs of dumb clients and typical Web publishing server implementations than DAV. That's the theory, anyhow. -Tim
If AtomPub, as defined, can do what you want, yes.On the other hand, I'm totally concerned with people who are trying to re-invent WebDAV in terms of AtomPub. For instance, using POST to implement namespace operations (operations, that after all give the client control over the URLs...) instead of just reusing COPY/MOVE.
Best regards, Julian