>p2p networks use central servers, torrents etc to help clients find >eachother). Then you get into storage/security issues when handily >private data.
These terms are a reminder of what is and will be important for that (XMPP-Wave-OT) systems, that's what people asking about when you say there is a new network uprising, actually ShareJS's Joseph are talking there about http://camlistore.org/talks/2011-05-07-Camlistore-Sao-Paolo for OT, and everywhere here and there 'networking' people are talking about decentralization and distribution, and from d-cent in DB to p2p in other parts, like a 1 step, there is a point that needs to be clearly realised - Wave with a systems like ShareJS are no more that old Wave but it's, still, with a bit of hacking you can run that ShareJS instance from your tablet or even phone for example, to organize home, small street network system and make it with your personal alias you can run that PyOfWave on your notebook to give a quick social for all your block/village (quick isn't about 10000 users like a booom), i can't say what Apache team proposed to done with WIAB, quickly, but it's certainly - not that kind of Wave system, so far, i think from what i'v seen - Apache Wave is more suits to upper corporative segment - that's where Java and all this structure comes to handy. that 'other' Wave systems, including amazing 'lightwave' (GO is a step forward in Wave, IMHO) are all built with Idea of quick weekend hacking on client/server staff, lightwave guy even has 'it's own' language for client writing) quickly customizable Cofeescript, html5 stack for clients, server languages that even a kid can understand and hack with, in a week, without big quality degrading But - it's obvious in that conception -- with that quick rising will be a quick downtime.. and that's the point where d-cent DB can help or p2p DB or some other -you know how many them now- DB persistence scheme, one can say - "that's all different servers, tech, staff, it's impossible to reuse from server to server without some super-migration!" but that's where the rightly done Federation miracle comes in.. or we fail to push it. >ConcurrentTreeColdStorage camp will be online soon, i'll try to ask him, i think they have an answer. here is - his latest opinion on federation >You could either relay BCP within XMPP stanzas, or invent your own >alternative for BCP using XMPP stanzas directly. Keep in mind that while >CTree can be relayed over just about any kind of wire or within any type of >container, it doesn't get along with other concurrency models (read: OT) >without some crazy confusing translation between CTree and the other model. >CTree simply doesn't work the same way as OT, which makes Wave federation >problematic but not impossible. - >p2p wont work to that end - it will only make writing a client more >complex. Also it was a kind of Win that even Google - still don't purged these GWave data, ain't it? you can imagine how many loses will be in that next gen social systems, so if we'll need p2p for stability (we can talk about Extensions distribution too, while that could be - almost fully a server-less html5 stack app, at least in PyOfWave conception) if we'll only lose stability or privacy, or ... by p2p than we need some other, better, solution, but solution, along with structuring and documenting all the OT and Wave achievements are needed anyway. That's why that piratepad exist now - to discuss with these servers parties from everywhere - all the possible solutions for our 'problems' along with making some Consortium Wiki or kind with FAQ and more advanced explanation of what's OT, Wave, XMPP,why it rocks and where the falls. Etherpad comes pretty handy in such situations, from history perspective. On Mon, Jun 13, 2011 at 3:18 PM, Thomas Wrobel <[email protected]> wrote: > ya - Looking over the pirate pad, I'm still not sure a p2p based > requirement for a wave system is going to be more simple to implement. > Scalability and path finding issues come into play. (even traditional > p2p networks use central servers, torrents etc to help clients find > eachother). Then you get into storage/security issues when handily > private data. > > Nathanael Abbotts ConcurrentTreeColdStorage > (https://docs.google.com/drawings/d/1iKiZJhtYBNrl8gm0A7rnjFyLwqBXtRvVFo6AJS0XcYw/edit?hl=en_GB&authkey=CPDB2IQH) > seems a neat solution for anominious (ie, no login) private access to > documents, but Im not sure how scalable it is. If you have 10,000 > documents and potentially thousands of participants, wouldn't it need > to check each key against each doc...on each client with the data? > And how would realtime updates work :? > How would you invite someone to a doc when they arnt online? (that is, > how would you get the "key" to them) > > Maybe I'm getting the wrong end of the stickwith how its working - as > I've said, I'll be more usefull client side then any server side > stuff. > Maybe other people with more knowledge of this stuff can weigh in here? > > Still, if the goal is to make wave easier to implement, I think adding > p2p wont work to that end - it will only make writing a client more > complex. > [/2 and a bit cents] > > The other thing...to everyone on this thread....is it possible to work > on client/server and sever/server protocols at the same time? If there > is really going to be big changes to the server side, wont this mess > up any client/server communications requirements? > I'm very keen to get a working c/s protocol, but at the same time, I > want to target my freetime where its more usefull and dont fancy > trying to (yet again) develop for another moving target thats going to > be useless in a few months. (as what happened to both FedOne and > PygoWave). > > -Thomas > [/keen to get on with actual coding] > > ~~~~~~ > Reviews of anything, by anyone; > www.rateoholic.co.uk > Please try out my new site and give feedback :) > > > > On 12 June 2011 23:56, ya knygar <[email protected]> wrote: >> Hello! >> >> according to the latest talks about rethinking the Federation protocols >> Were created a federation working group pad: >> >> http://piratepad.net/HET5ojzCXM >> >> briefly named - "Call for Participation in building a Federative >> XMPP-Wave-OT Social Networking Consortium" >> >> please, participate as much as possible in any of these fields, especially >> in a new protocol draft - reviewing that was discussed above. >> >> it highly depends on all interested Apache Wave members help, to make >> that Federation and, probably, Wave Consortium a reality, >> >> at least for Wave alike servers, we as a PyOfWave team are keen on it >> for a long time now, doing all what was possible to projects - follow >> that WaveProtocol.org standards and relate to Apache Wave as a >> reference implementation. We understand that - there was a busy - >> transferring and infrastructure uprising time - for WIAB, >> however - there is a high demand for changes in protocol, as you can >> see on that pad, we have done our best for creating a Wave Consortium >> community again, so, it's your turn to help. >> >> Thanks in advance. >> >
