jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ard Schrijvers (JIRA)" <j...@apache.org>
Subject [jira] Commented: (JCR-1201) Enable to load the indexing configuration from the repository with a config path like jcr:/location/in/my/workspace
Date Mon, 05 Nov 2007 12:35:52 GMT

    [ https://issues.apache.org/jira/browse/JCR-1201?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12540183

Ard Schrijvers commented on JCR-1201:

"What's your main use case; are your more interested in being able to dynamically reconfigure
things, or just in deploying the configuration in a war without having to put the configuration
in an external file? In the former case I think we need to carefully consider the implications,
but the latter case is much easier to solve. We could either merge indexing configuration
to the normal repository configuration file, or use a relative reference and load the configuration
from the classpath if that's where the repository configuration came from. "

The configuration from classpath I already implemented in an extended SearchIndex (which if
people like I can create a patch for). If the configuration starts with 'file:/'  it will
be taken from FS, otherwise from classpath. 

There are two usecases that currently are not possible (AFAICS)

1) have a different indexing_configuration per workspace
2) deploy a war, and tune the indexing_configuration through some interface which effectively
just changes repository content which holds the indexing configuration. Deploying a new war
for every project just to have a different indexing configuration seems cumbersome to me.

OTOH, I cannot like you judge the architectural implications ATM, so I might be wanting something
that does not fit into the current architecture. 

"Of course, in the long run I think we should go for a more general dependency injection model
for managing Jackrabbit configuration. This way you could dynamically inject whatever sort
of indexing behaviour you like, even one that does read the repository for the configuration."

You do think that in the future this kind of behavior is possible to fit in the architecture?
For now, I can work with the classpath solution, but in due time when external parties will
be working with our war, I would really like be able configuring the indexing without changing
the war. WDYT?

> Enable to load the indexing configuration from the repository with a config path like
> -------------------------------------------------------------------------------------------------------------------
>                 Key: JCR-1201
>                 URL: https://issues.apache.org/jira/browse/JCR-1201
>             Project: Jackrabbit
>          Issue Type: New Feature
>          Components: jackrabbit-core
>    Affects Versions: 1.3.3
>            Reporter: Ard Schrijvers
>            Priority: Minor
>             Fix For: 1.4
> ATM, you can provide an indexing configuration file via <SearchIndex> configuration
> <param name="indexingConfiguration" value="path-to-indexing-configuration-file"/>.

> Also see:
> 1) http://wiki.apache.org/jackrabbit/IndexingConfiguration
> 2) http://wiki.apache.org/jackrabbit/Search. 
> the 'path-to-indexing-configuration-file' currently must be a file in the file system.
Support for a indexing configuration which resides in the repository itself by enabling some
path protocol like jcr://location/in/workspace in the configuration. Specifically for war
deployment configuration files in the repository itself might be very useful.
> Perhaps adding an EventListener for instant re-configuration when the indexing configuration
changes might be nice as well, though some (most) changes in the indexing configuration might
require re-indexing, which would be bad to start automatically obviously.  

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message