On 11/8/05, Kyle Marvin <[EMAIL PROTECTED]> wrote: > On 11/8/05, Luke Arno <[EMAIL PROTECTED]> wrote: > > On 11/8/05, Kyle Marvin <[EMAIL PROTECTED]> wrote: > > > > > > Not that we need *more* options, but when we do introspection (whether > > > in core or elsewhere), I'm wondering why we need to invent any new XML > > > format. Why can't the introspection document just (itself) be an > > > Atom feed, > > > > This is what I keep saying and saying. > > > > I posted a pace last night. > > Yes, I think my informal proposal and your PACE end up at largely the > same place if you looked at what would happen on the wire. > > Your approach is probably more abstract, mine is probably more > incremental... which is beauty (or less controversial ;) is in the > eye of the beholder. If I were to author a PACE independently based > upon my description it would only replace Section 7 of draft-06. I > wasn't necessarily trying to purge collections but was trying to end > the format wars by offering a neutral (and hopefully agreeable) middle > ground. > > Thanks for the reference (and being persistent enough to help me get the > idea). >
Actually, I think your approach here builds on the the approach in my pace. You are using a feed of entries about 'collections' which is one use case that is possible within the approach described in my pace. I don't know if we should get that specific, though. I think the simplest thing that could possibly work is to define a rel type and use that to express relationships between the feeds and other resources. This way the feeds just describe themselves rather than having a feed which is for introspection with entries that describes other feeds which are for 'collections'. (Never mind that some collections are just views or gateways or what have you). In other words, I don't think we need an introspection resource at all (even one represented by a nice existing format like Atom or XHTML). All we need is discovery and let the resources describe themselves and each other as they will. - Luke
