I'd like to add my two cents in two parts. Earlier Peter claimed that JSON isn't extensible, but that is simply not truthful. If you define extensibility as having a warm fuzzy w3c document to tell you how to namespace things, well sure, JSON doesn't have any set rules or expectations there, but there is absolutely nothing to stop you from specifying your own extensibility rules in your particular use of JSON. You can name properties along the lines of 'com.creativepony.myThing' for a pretty good level of ownership, and require clients silently ignore unknown property names, as most already do. There are other techniques as well, but that is my favourite. If you want to up the performance in javascript, switch out the dots for underscores, to save yourself from using the slightly slower ["blah"] object property lookup operator.
That aside, I'm going against my usual reaction to all sights of XML as Data Format and supporting XMPP as the client protocol. It makes sense to keep everything as consistent as possible, share as much code as we can. By making use of XMPP as the client protocol, we can cut down on the amount of docs we need, as well as the volume of code needed in a compliant wave server. XML kind of sucks as a data format, but we're reasonably used to it now (we put up with the DOM when building the UI, so it's not that terrible to do so on the protocol end as well). Wave servers would also be able to implement their own JSON, RPC, or whatever protocols for talking to their own clients if they wish, or could just translate the XML DOM in to a quick and dirty JSON version for a small performance boost. — Jenna Fox On Oct 2, 9:48 am, elliottcable <[email protected]> wrote: > It looks like some people misunderstood the point… > > On Oct 1, 6:35 am, James Jones <[email protected]> wrote: > > > why not just keep using XMMP? > > Wave is *not* using XMPP right now for the C/S protocol. Only for > federation. The point of this discussion is to decide upon a C/S > protocol that doesn’t suck; I am personally lobbying for XMPP to be > said protocol. > > On Oct 1, 6:42 am, Peter Saint-Andre <[email protected]> wrote: > > > On 10/1/09 2:43 AM, elliottcable wrote: > > > I personally am interested in seeing XMPP become not only the > > > foundation of > > > the federation protocol, but of the C/S protocol as well. When I first > > > heard > > > about Wave, before getting into the Sandbox, I was very excited; it > > > sounded > > > like a great idea, based on great tools (XMPP!). Unfortunately, I was > > > extremely disappointed to find that XMPP really has nothing whatsoever > > > to do > > > with Wave, and I’d like to see that remedied. > > > I'd be happy to help with an XMPP binding. > > An XMPP ‘binding’? No, we don’t want to bind the existing RPC/protobuf > psuedo–protocol to XMPP, we want to replace it. Google has, multiple > times and places, said it’s very open to yanking out that hacky crap > and replacing it with a protocol that the community can agree on. So, > all that’s left is for us to agree on one, and tell google ‘implement > this.’ > > On Oct 1, 6:42 am, Peter Saint-Andre <[email protected]> wrote: > > > On 10/1/09 2:43 AM, elliottcable wrote: > > > Really, though, any sensible protocol that we can agree on would be > > > great. > > > Another option is possibly something involving JSON; that’d make > > > JavaScript > > > heavy–clients ridiculously easy to write for the web. > > > JSON is not extensible. IMHO that's going to make it difficult to build > > anything interesting. > > I disagree about JSON not being extensible, but it’s not really my > ‘favourite horse’ for this race, anyway. > > On Oct 1, 9:41 am, danbirlem <[email protected]> wrote: > > > XMPP seems like it should have been adopted and pushed for, cradle to > > grave when the words 'real time collaboration' were mentioned. > > However, I would not put it past that this protocol will rival XMPP in > > terms of flexibility. > > > Just my two cents... > > What do you mean by ‘this protocol’? The RPC/protobufs C/S protocol > currently in use? To even call that a protocol is kind of laughable. > Read above, it’s largely irrelevant as long as ‘we the people’ can > agree on a real protocol in a timely manner, so Google doesn’t change > their mind about being willing to use whatever we agree on. > > On Oct 1, 10:27 am, Peter Ferne <[email protected]> wrote: > > > On 1 Oct 2009, at 09:43, elliottcable wrote: > > > > I personally am interested in seeing XMPP become not only the > > > foundation of the federation protocol, but of the C/S protocol as > > > well. > > > Absolutely. It looks as if ProcessOne have already done some work > > towards this.http://tr.im/p1waveIsanybody from there on this list? > > It would be great if they were happy to share that work publicly. > > -- > > petef > > Hey, that looks pretty relevant. Maybe I’ll try to email some of their > guys with a Google Groups link to this thread later today. > > (Also, damn you, Google Groups, for automatically wrapping my text! > I’m so used to hard–wrapping everything I type at 78 characters… > sorry, to everybody else, for producing an original message unreadable > on Google Groups >_<) --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Wave Protocol" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/wave-protocol?hl=en -~----------~----~----~----~------~----~------~--~---
