Joseph. May be I am not getting it fully, I don't see how latency problem will be resolved in the P2P architecture you mentioned. The editor and the compiler are still remotely connected. Even if compiler has access to it's own OT system which builds the source files for it on every new delta, deitas will have to travel from editor node to compiler node. Assuming compile lies on server in a server-client system, and both editor and compiler have access to OT, its effectively the same process, server acting as remote node. On the other hand, editor, compiler, test-runners can still participate as agents, irrespective of the node they are at, possibly acting as clients, and do their respective jobs. P2P or not.
At max, the P2P federation gives flexibility of having different blocks on different nodes. But then, the more you distribute, the more you face the overheads of OT processing for each block. Not saying it will be huge, but a change from client-server for sure. Am I missing something? I have been working on a similar system, not specifically OT, but given replication as a necessity, I had to conclude client-server system has most practical advantages, including more control. On Tue, Jun 11, 2013 at 11:18 PM, Michael MacFadden < michael.macfad...@gmail.com> wrote: > Joseph, > > Very good commentary! I have drafted up some ideas on a hybrid system. > Actually I have seen two approaches. One uses a natively P2P protocol, > which then elects super nodes to act as "servers" in highly connected > clusters. At the other end of the extreme I have seen client server OT > use a P2P OT between federating servers. Both of these systems have pros > and cons. I do think that this has some benefit over pure P2P and pure > client server architectures. > > ~Michael > > On 6/11/13 6:44 PM, "Joseph Gentle" <jose...@gmail.com> wrote: > > >The biggest benefit to a P2P-capable system is federation. Currently, > >the wave federation algorithms create a distributed tree of servers, > >and they're vulnerable to netsplits if one of the root servers goes > >offline. Maintaining that tree is complex and unnecessary - there are > >better algorithms we can use instead. > > > >I imagine most devices will (most of the time) still connect to a > >single server. But if our OT system can manage P2P anyway and our > >system supports OT over arbitrary JSON blobs, there's a few really > >neat things we can do with that. > > > >So, this is something I've been wanting for ages - Imagine you're > >working on a software project. You put all the source code inside a > >JSON wave object and you edit it there. Because its a shared OT > >document, your compiler can join in on the document and annotate your > >code (live) with syntax highlighting, errors and autocomplete > >suggestions. It would know at a very granular level which functions > >need to be recompiled as you edit. Your unit tests could rerun > >themselves automatically and mark in your source code which tests > >failed and where. Its like a lego set for IDEs - your editor doesn't > >have to understand your programming language, your compiler doesn't > >have to know what editor you're using and where, and all the pieces > >can be on whatever computers are most convenient for you. And of > >course, you can pair program for free. > > > >You could build stuff like that on top of ShareJS today, but it would > >require a centralized server - and you have to put that somewhere. If > >the server is on your local machine, you can't upload the edits you're > >making to another OT server (like your private wave server). If its a > >remote server, you'll get terrible latency between making a change in > >your editor and your compiler finding out about it. With a system that > >supports P2P OT, we can make servers connect any way we want and It'll > >all just work. > > > >-J > > > >On Mon, Jun 10, 2013 at 1:38 PM, Dave <w...@glark.co.uk> wrote: > >> [John B - I wasn't sure where else it would be appropriate to ask this > >> question, but please forward on anywhere you think it appropriate] > >> > >> There are many things about Wave and WIAB that I would like to see > >>improved > >> / changed, but based on my readings I've been content with the TP1 OT > >> approach chosen by google (not that I'm even close to an expert) - even > >>if > >> the WIAB implementation would benefit from some love. > >> > >> But one of the things mentioned in the recent wave-forward hangout was > >>the > >> weakness in Wave's OT implementation for a required canonical version > >>of a > >> given wave (providing absolute ordering of changesets). Specifically, > >>this > >> effectively prevents 3 party P2P messaging where there isn't guaranteed > >>to > >> be that one canonical ordering. My understanding is that Joseph is > >>playing > >> with some alternative OT algorithms that are TP2, and therefore don't > >> require arbitration of changeset order. This was specifically called > >>out as > >> an advantage to support P2P messaging and running the full stack on a > >>phone. > >> > >> That got me thinking - why would you want to do that? What are the > >>benefits > >> of P2P messaging, and are there other reasons to need TP2? > >> > >> Most of the messaging and collaboration systems I could think of are > >> client-server (some with federated servers) and Wave/WIAB support this > >>with > >> TP1. Most networked phone apps that I'm aware of are also > >>client/server, > >> and at first glance this seems a good thing - it makes addressing > >>easier and > >> avoids issues with intermittent connectivity. The ability to have a > >>simple > >> "wavelike" server (and detached clients) > >> > >> I suspect I'm missing something, and I wondered if I'm alone? > >> > >> My understanding is that technical interop between the various wave-like > >> communities will need us to use the same OT alogrithm (eventually), so > >> clarity on the pros/cons of keeping or changing the wave OT approach > >>would > >> be a good first step in that direction! > >> > >> Dave > > >
