Funnelback 12.2.0 release notes

Released: 20 December 2012

12.2.0 - New Features

  • Support for multiple query processors.

  • Support for max_files_stored setting in web crawler Site Profiles.

  • A distinction is now made between user entered query parameters (meta_*, query*, query) and system generated ones (smeta*, squery_*, s). Spelling suggestions, blending and synonyms now applies only to the user entered parameters.

12.2.0 - Upgrade Issues

  • Groovy filters have been moved from SEARCH_HOME/lib/java/common to a specific folder SEARCH_HOME/lib/java/groovy.

  • The question.queryExpressions node of the the Modern UI data model has been removed. Query constraints generated from the meta_* date-related parameters are now stored in question.metaParameters. Form files or hook scripts that were using this node must be updated accordingly.

  • All the query processor options have now been converted to a standard syntax. All options are now specified in the form option=value and are consistent between command-line, configuration files and CGI parameters. External systems setting CGI values should be updated accordingly. See Query processor options (collection.cfg) for full details.

  • WARC storage is now the default for Filecopy collections. This will cause the Filecopy collections to be re-crawled entirely. store.raw-bytes.class (collection.cfg) can be set to com.funnelback.common.io.store.bytes.FlatFileStore to retain the previous behavior.

  • Query blending is incompatible with TAAT mode and is now ignored in that mode.

12.2.0 - Selected improvements and bugfixes

  • Text Mining performance improvements.

  • Text Mining is now compatible with meta collections.

  • Web Crawler:

  • Modern UI:

    • Support for Web resources folder on the Modern UI.

    • FreeMarker loop variables ..._has_next and ..._index are now exposed by Funnelback custom tags.

    • Use server hostname in query logs filenames.

    • Fix contextual navigation links containing clive parameters.

    • Fix click tracking links when there’s no user-entered query.

  • Indexer:

    • Fix handling of URLs containing multiple dots (e.g. http://server.com/...folder/file), URL encoding / decoding improvements.

    • Indexer: Fix host feature calculations on very large collections.

  • Query processor:

    • Improvements on partial matches in DAAT mode.

    • Overhauling the implementation of query blending so that the top result for the original query remains top of the blended list, unless query variants are given weights greater than or equal to 1.0.

    • Fix faceted navigation counts when using query blending.

    • Fix per-profile blending.cfg when run on the command line.

    • Fix handling of mandatory operators in DAAT mode.

    • Obey site.max_clusters=... parameter.

  • SSL support for Directory collections.

  • WARC files are now tied to their table-of-contents and cache companion files. An error will be raised if a WARC file is used with the wrong TOC or cache companion file.

  • Reduced number of background requests on the Admin home page to track collection updates progress.

  • Fix in directories permissions when running Funnelback under Apache.

  • Fix link in Trend_Alerts_reports emails.

  • Fix NTFS early binding DLS causing some results to never be returned to users.

  • ClearLocks mediator task now clears the stop flag as well as the update status.

  • Index merger: Fix merging indexes with same documents but different document flags.

  • Index orderer: Implement the same ranking defaults as in the query processor.

  • build_autoc: Provide a -partials option. This allows multi-word organic query completions to be triggered by sub-strings starting at each word. E.g. the suggestion australian workers union can be triggered when a person starts typing that string, or when they start typing workers union or just union.

  • build_autoc: Provide a -label_organics option for organic completions to be shown under the category General suggestions.