Tim Bray has a must-read lessons-learned/looking forward sort of post on Atom. How's this for a statistic: RFC 4287 took 17,944 email messages. Naturally I disagree with some of Tim's points. Notably - " RSS is Good Enough" - well ok dude, so why d'we need Atom?
I'll rhetoricise on that. To pick but one valid point from Tim's post is that the web needs to be writeable. RSS is read-only, hence not good enough. To pick another: RSS 2.0 can't reliably deliver content. To me the long-term killer follows from this - it can't reliably deliver data. Like my Web 2.0 mod'd Dalek likes to say: Deprecate!, Deprecate!
So...ok, GET and POST are everywhere (Lo-REST), few systems use PUT and DELETE (Hi-REST). But what do they use? Dunno, I guess what I've seen around most is ftp and variants. You establish a connection using a URI, you authenticate, you issue a method call, you maybe pass some content for a representation. Maybe Hi-REST hasn't been proven in practice like Lo-REST, but something very, very similar has. ftp methods are pretty explicit - closer to PUT/DELETE than burying everything in a POST.
But HTML's got a lot to answer for ;-) < plaintext> anyone?
(Hmm, just noticed Sam
ftp - I have to read on).
Mark Nottingham (aside from
a good question - what do we call these services/API things when
RESTifarians have a prior claim on the word
[uniform] "interface". Yeah, on reflection "services" or
"API"s aren't bad (Leigh's
and they're in common use. Got me trying to remember the reasons
the Atom API changed to the Atom Publishing Protocol...
A fly for the ointment:What makes a cool URI?
A cool URI is one which does not change.
What sorts of URI change?
URIs don't change: