Funnelback 14.2.0 release notes

New Content Auditor feature producing reports on a collection’s content.

Additional metadata classes with multi-character names are now supported.

Support for domain-level collection redirection.

Support for the tilde operator to up-weight specific query terms in daat mode.

Push collections are now able to provide parent meta collections with spelling and query suggestions.

Introduced a setting to control storage of URLs which have no content after filtering.

Version upgrades of a number of components, including the Java runtime (now version 8), the Groovy runtime (now 2.3), the Jetty web server (now 9.1).

Groovy Filters can now be placed under a @groovy/ folder under the collection configuration folder.

Access to the modern UI search interface via the admin web server port now requires authentication as it did prior to Funnelback version 12.

IP address restrictions are now consistently applied to all modern UI endpoints except log/redirect and search-session related endpoints.

Search session and history are now supported on non-web collections.

The groovy binary has changed location from tools/groovy-1.8.6/bin/groovy[.bat] to tools/groovy/bin/groovy[.bat] along with being upgraded to 2.3.7.

The blat binary has changed location from wbin/blat-2.6.2/full/blat.exe to wbin/blat/full/blat.exe along with being upgraded to 3.2.3.

When in count_urls mode, padre will no longer count URL protocols as a separate component. This does not affect any existing facet types, however the change would be visible in the raw padre data model.

Instant update on Filecopy collections will require valid URIs when deleting.

The mechanism for starting the Funnelback Jetty web server has changed from XML configuration files to Java code, with a Groovy script customization mechanism available for custom deployments.

If upgrading from version 14 you must run a bash script, supplied by the installer when run, before the installation can be upgraded.

The syntax for listing metadata classes has changed from a simple list of characters (e.g. xyz) to a comma separated list surrounded by square brackets (e.g. [x,y,z]) to support multi-character class names. Collection and profile configuration files are automatically updated during upgrade, however any external or custom configuration of such lists must be updated.

Funnelback now indexes metadata field boundaries by default (previously enabled only through the -ifb indexer option). A new -noifb option is available to disable this behavior if needed, however the change is not known or expected to affect any implementations.

crawler.store_empty_content_urls must now be set to true for any collection where URLs with empty content should be stored. This was never done prior to 13.2, and was always done in 13.2 and 14.0, but was not desirable in most cases, so a setting with a default of false has been introduced.

$CURRENT_VIEW in collection.cfg will now return live for instant update workflow and steps after swapping views, where as previously it returned offline. Any existing workflow steps may need updating to account for this.

Image scaling will now return .png images by default, instead of returning the type of the input image. When a different format is desired, this must be specified using the format cgi parameter. See Presenting Images for further details.

Connector collections, which relied on a third party framework which is no longer available, are no longer supported. Such collections should be removed before upgrading to 14.2.0 or later versions.

Push collections created prior to version 14.0.0 (also known as Continuous Updating collections) are no longer supported. Such collections should be upgraded to the new Push system or removed before upgrading to 14.2.0 or later versions. Contact Funnelback for assistance in migrating any existing collections of this type to the new push collection system.

Filecopy now stores properly encoded URLs (e.g. spaces encoded to %20, etc.). That will result in a re-crawl of all the documents with filenames (or containing folders) that needs encoding, causing the first crawl after the upgrade to take than usual as those documents will be re-filtered rather than being copied from the previous crawl.

Funnelback now internally uses a view URL parameter within the modern and classic UI systems. Implementations should avoid setting any value for this parameter.