search questions

[ChadDavis <ch...@gmail.com>]

1) Reading through the specification ( JCR 170 ) much of the search
related specifications are optional.  Does Jackrabbit implement most
of these optionals?

2) I must say, I'm a bit impressed with the level of search that you
get "out of the box".  Big step up functionally from my having to
managing the dumping and indexing of objects stored in the db, etc.
Is the XPath interface to search in the Jackrabbit performant?  Even
with the remote server deployment?

Re: search questions

[Ian Boston <ie...@tfd.co.uk>]

On 8 Nov 2009, at 08:16, ChadDavis wrote:

> 1) Reading through the specification ( JCR 170 ) much of the search
> related specifications are optional.  Does Jackrabbit implement most
> of these optionals?

AFAIK, all

>
> 2) I must say, I'm a bit impressed with the level of search that you
> get "out of the box".  Big step up functionally from my having to
> managing the dumping and indexing of objects stored in the db, etc.
> Is the XPath interface to search in the Jackrabbit performant?

AFAIK, Yes, but ultimately it depends on a number of things.
Size of repository, billions of items will generate a larger index of  
pointers to documents.
The larger the size of term vocabulary will create a larger inverted  
index, and if you need to sort then that may require more memory.


>  Even
> with the remote server deployment?

Performance over RMI is heavily dependent on the latency of each call  
and the frequency. Single calls are performant, if the application  
makes 1000's of calls to achieve a single user operation, then network  
latency will become dominant.

HTH
Ian