Mailing-List: contact dev-help@jackrabbit.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@jackrabbit.apache.org
Received-SPF: pass (nike.apache.org: domain of stefan.guggisberg@gmail.com
 designates 209.85.210.170 as permitted sender)
MIME-Version: 1.0
Date: Thu, 29 Dec 2011 21:49:39 +0100
Message-ID: 
 <CAFYk8Nm3aSuc=9PRwQBWozj99rQGgnygot41nr1BFM70nRRR2g@mail.gmail.com>
Subject: Re: [jr3 microkernel] New transient space implementation
From: Stefan Guggisberg <stefan.guggisberg@gmail.com>
To: dev@jackrabbit.apache.org
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

On Thu, Dec 29, 2011 at 6:19 PM, Michael D=FCrig <michid@gmail.com> wrote:
> Hi,
>
> Over at the sandbox [1] I implemented a new (AFAIK) approach for the JCR
> transient space. The gist of it is in the ChangeTree class [2] while some
> more Microkernel specific things are factored out into the TransientSpace
> class [3].

great work! it's good to see a fresh take on jcr client implementation.

some comments from the peanut gallery follow inline ;)

>
> The key idea is to only store changes to the node hierarchy in the
> ChangeTree and to do that in a way which allows reconstructing a list of
> corresponding operations (add, remove, move for nodes and set for
> properties) from that ChangeTree. This differs from other implementations
> where the list of operations is kept *together* with the modified hierarc=
hy.
> Furthermore with the ChangeTree approach no untouched nodes need to be ke=
pt
> in memory.

that's also true for the existing jackrabbit-core implementation.

> Finally the list of operations reconstructed from a ChangeTree is
> minimal. That is, all cancelling operations are reduced as much as possib=
le:

same here, jackrabbit-core does consolidate transient changes as well.

cheers
stefan

> add followed by remove of the same item results in no operations, add
> followed by move of the same node results in the node being added at the
> target of the move, move followed by remove of a node results in the node
> being removed at the source of the move and move followed by another move=
 of
> a node results in a single move from the original source to the eventual
> target. More details are in the class comment of ChangeTree and even more
> details in the implementation ;-)
>
> Michael
>
>
> [1] http://svn.apache.org/viewvc/jackrabbit/sandbox/jackrabbit-microkerne=
l/
>
> [2]
> http://svn.apache.org/viewvc/jackrabbit/sandbox/jackrabbit-microkernel/sr=
c/main/java/org/apache/jackrabbit/state/ChangeTree.java?view=3Dmarkup
>
> [3]
> http://svn.apache.org/viewvc/jackrabbit/sandbox/jackrabbit-microkernel/sr=
c/main/java/org/apache/jackrabbit/state/TransientSpace.java?view=3Dmarkup