Funnelback 12.0.0 release notes
Released: 30 August 2012
12.0.0 - New Features
-
Major improvements to Filecopy and TRIM collection types - Performance, reliability and the ability to search collections during gathering.
-
Text Mining: Extraction of named entities and definitions from text.
-
Simplified support for managing Funnelback query processor systems attached to an administration server.
-
Graphical redesign of the Funnelback administration interface.
-
Improved query processing performance with skip-block index structure.
-
Improved default ranking parameter settings.
-
Automated document classification.
-
Document level security and metadata integration with Squiz Matrix.
-
Date based facet generation.
-
Image presentation tools including scaling and web page preview rendering.
-
Support for NOARCHIVE and NOSNIPPET directives in robots metatags.
-
Significant improvements to support for CJKT languages (including support for mapping of traditional to simplified Chinese and vice-versa).
-
Support for localisation of search template files.
12.0.0 - Upgrade Issues
-
Due to the standardization of many query processor options the format of some CGI parameters has been changed, and must be updated in any external systems setting CGI values. See Query processor options (collection.cfg) for full details.
-
The Redis service included with Funnelback 11.x may not shut down correctly during upgrades. Please ensure you shut down Funnelback manually before upgrading and check that the redis-server process is stopped.
-
Funnelback 11.x and greater embed a 64-bit Java virtual machine (JVM). If you are upgrading from an older version of Funnelback you may need to increase the max_heap_size by 50% to account for the increased memory requirements of the 64-bit architecture.
-
Some components of Funnelback now depend on the installation of freetype and fontconfig on the local system. On RedHat/CentOS systems, these may be installed with
yum install freetype fontconfig
if they are not present. -
The
<@IncludeUrl />
Modern UI tag is now part of thefb
namespace:<@fb.IncludeUrl />
. Form files must be updated accordingly. -
The default filter for PDF to HTML conversion is now Tika (previously
pdf2html
). This might slightly affect the rendering of converted PDF files. -
The
-cjkt
query processor option is no longer required and no longer supported. -
The truncation operator (
*
) is no longer supported in queries unless the-service_volume=low
query processor option is used.
12.0.0 - Selected improvements and bugfixes
-
Standardized the syntax of many query processor option settings.
-
Improved query completion performance.
-
Suppress duplicate descriptions in query completions.
-
Ability to bias result ranking based on gscope settings.
-
Funnelback shell tool for manual manipulation of stored content.
-
Services for remotely performing storage and mediator operations on Funnelback servers.
-
The Modern UIData Model
rawInputParameters
andinputParameterMap
are now backed by the same Map: Any change made in one will be visible in the other. Existing hook scripts and FreeMarker template shouldn’t be affected but some code simplification might be possible when parameters were injected into those Maps. Additionally, theadditionalParameters
map now takes arrays of Strings as values, as opposed to just Strings previously. -
padre-sw
may now be accessed directly at/s/padre-sw.cgi
for debugging purposes. -
Support for selectable metadata values.
-
Ability to turn off stemming of particular words when stemming is on by default.
-
More accurate reporting of filetypes.
-
Facet counts are now supported (through estimation) in DAAT mode.
-
Support many more complex query types in DAAT mode.
-
Support for anchor-text processing and redirect/duplicate information in push collections.
-
Ability to restart gathering from a checkpoint in Filecopy and TRIM collections.
-
Support for individual server authentication details via site profiles.
-
Fixed redis server security so manual security configuration is no longer required.
-
Improved URL status collection tool for web collections.
-
Added ability to snapshot push collection indexes so they can be distributed to remote servers safely.
-
Improved Funnelback uninstaller to remove environment settings.
-
Funnelback now includes a full Java development environment (rather than just the runtime environment) to provide better performance monitoring tools.
-
Fixed incorrect spelling in continuous updating configuration parameter names.
-
Fixed restarting of push collections after a service restart.
-
Improved logging for jetty web applications.