Mailing-List: contact user-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@couchdb.apache.org
Received-SPF: pass (athena.apache.org: domain of jens@couchbase.com designates
 206.225.164.30 as permitted sender)
From: Jens Alfke <jens@couchbase.com>
To: "user@couchdb.apache.org" <user@couchdb.apache.org>
Date: Tue, 5 Nov 2013 14:49:51 -0800
Subject: Re: CouchDB vs RethinkDB
Thread-Topic: CouchDB vs RethinkDB
Thread-Index: Ac7aeVdKcweA5MCVTfCnJcuqed2/vQ==
Message-ID: <A2A9E11A-AED7-442D-89E5-7145027D2A78@couchbase.com>
References: <BC8C6854-B37E-49C8-A160-EEAB285EE66E@sri.com>
 <CAMGHjAQtEfg7EbPWAEqH1+pEtZEW-Dktoza3mWQ1wHLC8=++Tw@mail.gmail.com>
 <CAOA1H+JdUby-y0RdozFANnyQSc1wjtTdup9pkyP=KuG4y7aT8w@mail.gmail.com>
 <1C1EDFED-1208-496D-AFD3-220ED5A8400E@sri.com>
In-Reply-To: <1C1EDFED-1208-496D-AFD3-220ED5A8400E@sri.com>
Accept-Language: en-US
Content-Language: en-US
acceptlanguage: en-US
Content-Type: text/plain; charset="Windows-1252"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0


On Nov 5, 2013, at 1:13 PM, Jim Klo <jim.klo@sri.com> wrote:

> It seems to me that one could build an add-on to any database to support =
this I think? =20

Yes, but you need to add a significant amount of metadata to each document/=
record/row since you need to track its revision tree[1].
(I am currently building exactly such an add-on so I know whereof I speak :=
)

> I was actually wondering how difficult it would be to build a 'generic re=
plication api' that leverages the same CouchDB replication protocol=85  Has=
 anyone endeavored to try anything like this?  It seems like it should be s=
traight forward.

There are a couple of re-implementations of the CouchDB replication protoco=
l: PouchDB and Couchbase Lite (in both Objective-C and Java implementations=
).

I have my own documentation of the replication algorithm[2] and there=92s a=
nother description that=92s recently been added to the CouchDB docs.

I don=92t know about =93straightforward=94 =85 it=92s not rocket science, e=
specially now that there is actual documentation of the protocol, but it=92=
s a fairly complex engineering task, especially the failure handling.

=97Jens

[1] Gory details in the 2003 paper "The Hash History Approach for Reconcili=
ng Mutual Inconsistency=94: <http://citeseerx.ist.psu.edu/viewdoc/download?=
doi=3D10.1.1.12.5482&rep=3Drep1&type=3Dpdf>. I don=92t know for sure, but I=
 suspect this paper was an influence on the design of CouchDB since what it=
 describes is very similar to the way CouchDB replication works.
[2] https://github.com/couchbase/couchbase-lite-ios/wiki/Replication-Algori=
thm