- From: T.V Raman <raman@google.com>
- Date: Wed, 9 Mar 2011 16:57:17 -0800
- To: nathan@webr3.org
- Cc: raman@google.com, nrm@arcanedomain.com, www-tag@w3.org, jeni@jenitennison.com
1+ -- articulating this in terms of continuations is illuminating. Nathan writes: > T.V Raman wrote: > > Some thoughts from the past: > > > > ? and # are symmetric -- ? is server-side # is client-side. One > > of the inconsistencies in web-arch as defined is that it perhaps > > fail to recognize this symmetry fully. > > That sentence alone is very clarifying :) - and as Jeni noted, > developers are already using the two symmetrically, using paths and > key=value pairs in fragments. > > > Secondly, I believe there is a continuum from documents to > > applications, an application with 0 interactivity is perhaps a > > document. Another way to look at this is: > > > > "The Document Is The Interface" but "The Interface Is Not A > > Document" --- An interface is actually a collection of > > documents, where each document snapshots a particular state in > > the overall app;ication interaction. > > This is a very interesting line of thought, so we could see the new > twitter site (for example) as providing a document which may have > potentially infinite states, and the server as providing a snapshot of > that document in a state from which the others can be composed by the > client (a continuation even), and fragments as being references to one > of those composed states. > > This would be a very useful and efficient pattern for clients, servers > and the network; indeed without it each possible document state would > need to be generated, stored and then referenced and cached around the > web, which is almost infeasible. > > I guess in to this comes references to composite resource state (as with > the hash-bang cases and Ashok's hashInURIs write up). > > One could say that both servers and clients now have an equal role in > composing and controlling resource state, with resource representations > over the wire being continuations. Perhaps it could be suggested that > the universal interface works in both directions, hiding the > implementation details on both sides, exposing only the interface and > using resource identifiers and continuations with a well known media > type to transfer state between components... > > I'll stop here, but interesting to say the least.. > > Best, > > Nathan > > > Noah Mendelsohn writes: > > > I've posted some thoughts at [1] on the identification of documents in Web > > > applications. This relates to the draft Ashok is editing on client-side > > > URIs [2](titled: "Repurposing the Hash Sign for the New Web", though one of > > > the points in my posting is that we need to look at "?" as well as "#", so > > > I'd really like to change that title.) > > > > > > The posting also relates to the hash-bang #! controversy, and I'll be > > > scheduling a discussion of that for Thursday. Please try to read both my > > > posting [1] and Jeni's [3] before this week's call. Last I heard from Amy, > > > Tim will be joining us. I expect to do the agenda tomorrow morning. Thank you. > > > > > > Noah > > > > > > [1] > > > http://blog.arcanedomain.com/2011/03/identifying-documents-in-web-applications/ > > > [2] http://www.w3.org/2001/tag/2011/02/HashInURI-20110208 > > > [3] http://www.jenitennison.com/blog/node/154 > > --
Received on Thursday, 10 March 2011 01:05:38 UTC