Hi,
Thanks for your comments. I should have read the wiki more carefully ;o)
I thought that references allowed me to better express my relations (so
that users of my data structure knew what to put where).
As for your solution with soft references, I will consider it as it is
closer to what I want ;o).
I have not invested enought time in the event part of the JSR and
thought that references was the way ;o).
I'll look into it as soon as possible.
Thanks,
Emmanuel
Alexander Klimetschek a écrit :
If you have lots of references in your system, maybe you could
consider using "soft" references by storing the reference in a PATH
property with simply the path to the property. Also have a look at the
content modeling tips on the Jackrabbit wiki, which explains why
references in JCR are not recommended:
http://wiki.apache.org/jackrabbit/DavidsModel#head-ed794ec9f4f716b3e53548be6dd91b23e5dd3f3a
You no longer have the getReferences() method available, but initially
you could build a simple search a la
//element(*, my:referencers)[EMAIL PROTECTED]
Then iterate over the @ref properties using the getRows() iterator and
do a query for each @ref and build a hash map of all documents that
are referenced. Then you install an observation listener for all
"my:referencers" nodes (that have references) and update the hash set
when a new reference is added or removed. This way you have a quick
lookup in your hash map for nodes that are referenced.
Regards,
Alex
On Fri, Jul 18, 2008 at 7:46 AM, Emmanuel Hugonnet <[EMAIL PROTECTED]> wrote:
Alexander Klimetschek a écrit :
On Thu, Jul 17, 2008 at 4:44 PM, Emmanuel Hugonnet <[EMAIL PROTECTED]>
wrote:
I would like to get all the nodes of a list that are NOT referenced.
Is there a way to do this with a XPath Query because it looks like
reference is a one way axis for xpath :( .
I think it is not possible with a query. You can iterate over all the
nodes you want to inspect, call getReferences() on it and hence find
out, which ones are referenced by another node.
Regards,
Alex
Thanks,
That's what I was doing but with over 3000 nodes this is quite a bottleneck
for performance :-(
Maybe I should add an attribute for better filtering but I am moving from
the data to the logic :-(
Regards,
Emmanuel
