Funnelback 13.0.0 release notes

Released: 31 October 2013

13.0.0 - New Features

  • Social media gathering support

  • Curator search result page customization

  • Recommendations system

  • Improved, responsive default form

  • Modern UI search session and history support

  • Various Modern UI improvements

    • Cache support with better display customization

    • Click tracking with tracking of auto-complete actions

    • Document level security support for all collection types

  • Result collapsing

  • Support for the Manifold-CF Enterprise Connector framework (with many included repository types)

  • Support for Windows Server 2012

13.0.0 - Selected improvements and bugfixes

  • New custom collection type supporting custom-developed gathering components

  • New -filepath_exclusion_pattern indexer option to exclude files from the index before they’re scanned

  • A new RSS FreeMarker template has been created for the Modern UI

  • Mediator.pl is now capable of pushing report folders

  • Better support for workflow scripts written in Groovy by expanding the $GROOVY_COMMAND variable

  • Gscopes can be used with Push collections

  • Query Independent Evidence (QIE) can be used with Push Collections

  • Ability for autocompletion to extend the existing query (e.g. to select facets)

  • The bundled versions of Java, ActivePerl and Groovy have been upgraded

  • A range of improvements/refinements to the Trimpush gathering process

  • Data API allowing access to Funnelback-stored data for a given URL

  • Fixed bugs relating to tuning when the collection and profile both specified query processor options

  • Ability to enable geolocation of user’s IP (Using MaxMind GeoLite) address to allow result customization based on location

  • Improvements to in-crawl form interaction

  • Performance improvement when using the fb.IncludeUrl tag with start and end restrictions

  • Support for grouping collections within Funnelback’s administration interface

  • New filter step (InjectNoIndexFilterProvider) for marking content as no-index during filtering

  • Compression of archived search log files is now performed automatically

  • Improved calculation of facet count estimates in document-at-a-time query processing

  • Validation and concatenation of external metadata files

  • Support for conversion of WARC files containing non-web URLs to file and directory layouts

  • Modern UI controllers for serving repository documents

  • Access restrictions based on X-Forwarded-For headers and CIDR notation

  • Correct logging of request source when multiple X-Forwarded-For entries are provided

  • Replaced file editor component avoiding cut/paste issues previously experienced

  • Ability to clear update locks from within Funnelback’s administration interface

  • Relative date formatting in search form files

13.0.0 - Upgrade Issues

  • As of version 13 Funnelback no longer supports Windows Server 2003 or any 32 bit operating system.

    • If you need to continue using Classic UI under IIS for document level security, you need to install a separate 32bit copy of ActivePerl and configure IIS accordingly.

  • Funnelback no longer supports the creation of new classic UI collections (though existing ones are still supported on upgrade)

  • The field userIdToLog in the Modern UI data model has been renamed to requestId. Hook scripts must be updated accordingly.

  • The data model node extra containing extra searches has been renamed to extraSearches in the JSON and FreeMarker mode to be consistent with the name used in hook scripts and the XML output. Existing forms must be updated accordingly. Backwards compatibility can be achieved by setting <#assign extra=extraSearches? /> at the top of the form file.

  • The post_update_command is now run in the background on Windows (as of Linux). That might affect scripts that were relying on the exit status of the this command to fail the update.

  • Handling of URLs logged with unencoded spaces within analytics has been improved. Collections with affected analytics databases should remove the old database after upgrading to have the database fully rebuilt.

  • PushIndex in mediator.pl has been deprecated, PushView should be used instead.

  • The ActivePerl installation folder no longer contains a version (e.g. $SEARCH_HOME/linbin/ActivePerl-5.12 is now $SEARCH_HOME/linbin/ActivePerl). Custom scripts that accessed ActivePerl at the old location must be updated accordingly.

  • The RedisDB Perl module is not included with Funnelback anymore, but is now installed in ActivePerl. Scripts using a different perl interpreter (such as the OS perl) will need to install this module.

  • The HTML markup for query completion categories has changed from <h2> to <li class="ui-autocomplete-category"> to comply with jQuery UI best practices. CSS must be updated accordingly.

  • A new 'databases' directory has been added to a collection’s 'live' and 'offline' directories. This is used for storing database files required by some features e.g. Recommendations and Text Mining. The 'update-configs.pl' script run during upgrades will create these directories for existing collections and move TextMiner database files if required.

  • When using multiple query processors you must change when publish_index.pl is called. It will now push from live to offline, so must be called during the post_archive_command rather than the post_index_command.

  • The post_update_command to support record types and classification security on TRIMPush collections has changed. It was previously taking the server details and path to the index folder, but is now taking the name of the collection and will automatically look up the relevant settings from collection.cfg