jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jukka Zitting" <jukka.zitt...@gmail.com>
Subject Re: svn commit: r504623 - /jackrabbit/trunk/jackrabbit-core/src/main/java/org/apache/jackrabbit/core/SessionImpl.java
Date Thu, 08 Feb 2007 09:21:43 GMT
Hi,

On 2/8/07, Stefan Guggisberg <stefan.guggisberg@gmail.com> wrote:
> On 2/7/07, Jukka Zitting <jukka.zitting@gmail.com> wrote:
> > Sounds reasonable, but I'm not convinced that the performance gain is
> > noticeable enough to justify the change. What's the access pattern
> > we're trying to speed up?
>
> well, applications that make *heavy* use of  Node.getNodeByUUID do
> benefit considerably as shown by our profiling data.

My assumption is that the isNodeType() call should be pretty fast
since the NodeState with the type information has already been loaded.
Perhaps there's a performance issue with the EffectiveNodeType
handling?

I would assume that an application that uses getNodeByUUID() will also
do something with the returned Node. This change obviously speeds up
the getNodeByUUID() method call especially if the NodeState is already
cached, but is the speedup big enough to have a noticeable effect on
any composite operation? Intuitively I'd assume isNodeType() to be
faster than things like getNode() or getProperty(). If this assumption
is true, then removing the call would only have a marginal effect on
the overall performance.

BR,

Jukka Zitting

Mime
View raw message