[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
HTTP, WebDAV, and POST
Very good descriptions by Mark and John! I wanted to add that it may
not have to be an either/or situation, and there are advantages to
both techniques. Just a few notes:
* As noted by Julian, for this first round of Atom specs we don't
need to "use WebDAV", we can specify Atom usage of basic HTTP
methods that are inherently upwardly compatible with WebDAV.
* Eventually, Atom will grow up to need robustness or collaborative
features like transactions, locking, ACLs, etc. This is the scope
and purpose of WebDAV, so assuming Atom would eventually add these
operations "into its API" would be unnecessary and conflicting
work. The scope of work Atom can focus on can be much smaller:
setting the context of using WebDAV for journal and chronological
resources.
* I'm already somewhat concerned about Atom developing a new, unique
way of associating metadata with arbitrary web resources (when
people talk of "posting metadata after they post their cat
picture"). WebDAV already does support that. I'm not concerned
about this in the context of textual Atom entries, though, as
entries are complete resources in-and-of-themselves.
* Although nominally a "web resource", a large number of people seem
not to think of entries managed through the Atom API as actionable
web resources, they seem to think of them like XML records
manipulated through the Atom API. This perspective may hinder
eventually using WebDAV facilities like transactions, locking, and
ACLs on entries themselves.
* Although WebDAV does not use POST, there does not seem to be any
conflict in Atom specifying POST for creating server-located
resources and at some point using WebDAV actions on the resources
after they've been created.
* Additionally supporting PUT-to-create for arbitrary resources (in
the current discussion), and for entries (separate discussion)
positions Atom to be upwardly compatible with WebDAV.
-- Ken