Distributed version control?

[arauschmayer]

I've blogged about requirements for social applications and RDF
engines [1]. Can Open Anzo do distributed version control (including
peer-to-peer synchronization)? If not, then is something like this
planned?

Thanks!

Axel

[1] http://2ality.blogspot.com/2009/07/requirements-rdf-and-social.html

[Lee Feigenbaum]

Hi Axel,

Open Anzo's replication service currently does 2-way synchronization
between Anzo clients and an Anzo server, but (as you've noted) it does
not do distributed peer-to-peer synchronization (in the mannger of
something like hg or git). I'm guessing that distributed version control
could be implemented on top of replication, but to take the central Open
Anzo server out of the picture would be a substantial undertaking and
one we're not planning on currently.

Lee

--
Lee Feigenbaum
VP Technology & Standards, Cambridge Semantics Inc
l...@cambridgesemantics.com
1-617-553-1060

[Sean Martin]

Hello Axel,
I am not sure how active the project is these days, but it might be worth
your while looking at the DBin project http://www.dbin.org/overview.php I
thought it was interesting work in the area when I looked at it quite a
while back. Paper @ http://www.dbin.org/dbinoverviewiswc2006.pdf


Kindest regards, Sean


--
Sean Martin,
President & CTO, Cambridge Semantics Inc.
email/XMPP: se...@cambridgesemantics.com
phone: +1 617 606 3411
FOAF: http://www.cambridgesemantics.com/people/sean

[arauschmayer]

> Open Anzo's replication service currently does 2-way synchronization
> between Anzo clients and an Anzo server, but (as you've noted) it does
> not do distributed peer-to-peer synchronization (in the mannger of
> something like hg or git). I'm guessing that distributed version control
> could be implemented on top of replication, but to take the central Open
> Anzo server out of the picture would be a substantial undertaking and
> one we're not planning on currently.

Makes sense. That means that two servers cannot synchronize, either,
right?

If a client synchronizes with a server, does it always have to be the
same one? If synchronization is based on some kind of history, I would
suspect so.

Still, Open Anzo gets a lot of things right: it recognizes what
features are needed by many multi-user applications, while not being
too heavy-weight.

[Lee Feigenbaum]

arauschmayer wrote:
>> Open Anzo's replication service currently does 2-way synchronization
>> between Anzo clients and an Anzo server, but (as you've noted) it does
>> not do distributed peer-to-peer synchronization (in the mannger of
>> something like hg or git). I'm guessing that distributed version control
>> could be implemented on top of replication, but to take the central Open
>> Anzo server out of the picture would be a substantial undertaking and
>> one we're not planning on currently.
>
> Makes sense. That means that two servers cannot synchronize, either,
> right?

That's right, for now, though this is something that we are planning to
look at down the road.

> If a client synchronizes with a server, does it always have to be the
> same one? If synchronization is based on some kind of history, I would
> suspect so.

Multiple clients can all synch with the server, but they are all
separate, distinct replication operations, that maintain their own
independent sync points (replica states, effectively).

> Still, Open Anzo gets a lot of things right: it recognizes what
> features are needed by many multi-user applications, while not being
> too heavy-weight.

We hope so, that's what we're aiming for - always happy to help out
however we can.

Lee