Hi,
I haven't touched the workspace configuration file.
After upgrading an exisiting repository (by replacing all the jar's +
adding dependencies), and removing it's index files, the behaviour all
the 'old' files can't be searched through either.
I must be missing something really simple, or I did muck up my config
somewhere before the upgrade.
Regards,
Jop
Marcel Reutegger wrote:
Hi,
did you make sure that the workspace configuration file (workspace.xml) is the
same as the one you used with jackrabbit 1.4.5? that's where the text extractor
classes are configured, which are responsible for extracting the text from your
pdf and odt files.
regards
marcel
Jop Zinkweg - Initworks B.V. wrote:
Hello everyone,
After upgrading from Jackrabbit 1.4(.5) to 1.5(.0), the behaviour of our
search box changed.
Previously we were able to search the 'binary' contents and additional
properties of a jcr:content node using the following query:
SELECT * FROM bos:correspondentie WHERE CONTAINS(.,'abc')
After upgrading this query only returns 'old' files (uploaded while
running 1.4) which have 'abc' in them (pdf / odt files).
When searching for a value known to be in a 'meta' property both 'old'
and 'new' (1.5) files are returned.
After removing the index both the 'old' and 'new' files can only be
found using their properties.
This leads me to believe the indexing behaviour (and not the query
behaviour) has changed between 1.4 and 1.5.
We're running a vanilla 1.4 configuration, and looking at the 1.5
vanilla config nothing has changed in respect to the searching/indexing
default setup.
Our node structure is as follows:
* 'folder' = nt:folder
* 'file.odt' = nt:file
* 'jcr:content' = nt:unstructured (+ bos:correspondentie mixin
defining some properties)
Has the indexing behaviour changed in 1.5, or am I looking at another
problem entirely?
Thanks in advance,
Jop Zinkweg
