You are here

Block title
Block content

NOW AVAILABLE: Fedora GSearch Version 2.4.1

From Gert Schmeltz Pedersen, Special Consultant, Technical Information Center of Denmark

We are pleased to announce the release of Fedora GSearch version 2.4.1. The primary feature of GSearch is that it makes it relatively easy to make your digital contents in Fedora searchable for yourself and your end-users.

Download it here: .

The source code is here: .

The full list of GSearch issues is here: .

It is the second and final step of an activity that was initiated at OR2011 in June (, which also has an introduction to GSearch functionality. 

The first step was the release of GSearch 2.3 in September.

Where 2.3 focused on the out-of-the-box user, making it easier to make Fedora objects searchable, this version focuses on more complex use cases, besides upgrading to the latest versions 3.5.0 of Apache Lucene and Apache Solr. Briefly, highlights of this release are:

• Extraction of text and metadata from datastreams using Apache Tika, which means that you can index almost any format. You can also index metadata extracted from non-text datastreams like images. Thanks to Adam Soroka and Serhiy Polyakov.

• Customizable end-user search client for GSearch. Thanks to Christian Orthmann, DTU Library.

• Elaboration and exemplification of filtering of search results by access constraints. Thanks to Swithun Crowe.

• Management of GSearch configurations in Fedora objects. Thanks to Adam Soroka. The current "solution" may not be, what he wanted, improvements may be requested.

• Many-to-many relationship between index documents and Fedora datastreams and/or objects. Thanks to Jonathan Green for initiating it.

• Interaction with the Resource Index, using a mechanism called "embedded queries", allowing you to embed risearch queries in Lucene or Solr queries, and vice versa.

• Performance measurements report, thanks to Morten Sørensen, DTU Library. Measurements taken using Apache JMeter, on a production quality platform, giving some insight into the performance implications of various choices.

• Improved documentation.

Many thanks to DEFF, Denmark's Electronic Research Library (, and to DTU Library (, for funding this development and the previous developments starting in 2006.