Funnelback 6.0.0 release notes
Released: 31 Oct 2006
New features
-
Completely rewritten administration interface, including the ability to create multiple users, and delegate permissions for specific tasks and collections.
-
New crawler for TRIM Context repositories. (
http://www.towersoft.com/global/Product/TRIM+Context+6+
) -
Per-document security support over TRIM and file-copy (NTFS only) collections.
-
Higher performance search query reporting system with many new reports and plug-in architecture.
-
Inbuilt support for incorporating end-user click data to improve search result ranking.
-
Internal indexing and query processing rewritten to work internally in UTF-8 for better foreign character set support.
-
Thesaurus / query expansion.
-
Ability to eliminate words known to be misspelled in the corpus from the corpus dictionary.
-
Direct access to search results with spelling suggestions via
xml.cgi
REST web service. -
General scoping mechanism for setting and querying bits on specific documents.
-
Sorting options expanded to cover all metadata fields.
-
padre-fl
tool for removing documents from search indexes is now included by default
Other changes
Required environment changes
-
Perl 5.8+ is now required (up from 5.6)
-
Java 1.5+ is now required (up from 1.4.2)
-
libiconv is now required
-
Windows 2000 is no longer supported
-
Under Linux, glibc 2.3.3+ is now required (up from 2.2.5)
Functional changes
-
System name changed from Panoptic to Funnelback. Includes changes to default forms and license key location.
-
Under windows, the default administrator user is now named
funadmin
rather thanpanadmin
. -
Default search forms now include CSS styles in-line for easier customization.
-
Default search forms no longer provide a link to user help documentation. Sample user documentation is still available within the administration documentation.
-
Default search results per page changed from twenty to ten.
-
Spelling suggestions are now provided by default for the following CGI parameters:
query
,query_and
,query_or
,query_phrase
andquery_sand
. Customization is available via thequery_parameters
attribute of theCheckSpelling
tag. -
refine.cgi now processes all
query_*
CGI parameters except query_not. Previously, only the query was processed. -
Featured pages are now matched against the search query prior to canonicalization, rather than after.
-
When configuring Apache, Funnelback now binds to all IP addresses (*) rather than the old behaviour of choosing the first IP address of the first interface.
-
During installation, addition of the web server user to the search group is now optional.
Bug fixes
-
Crawler check-pointing previously failed under Windows
-
Crawled document URLs are now canonicalized more effectively.
-
Dates in the future can now be indexed, where as previously they were ignored.
-
Fixed default page matching in
external_metadata.cfg
-
Installer no longer assumes that web server user and group are the same.
-
analyse-size.pl
previously displayed warnings when run in empty directories. -
crawl.pl
previously produced warnings if log, idx or data directory was already empty.
Upgrade issues
-
When upgrading an Apache configuration, the old
httpd-panoptic.conf
configuration file will be removed from Apache’shttpd.conf
, and replaced by a newhttpd-funnelback.conf
configuration file. Any environment specific customization in thehttpd-panoptic.conf
file should be manually transferred to the new configuration file during the upgrade. -
Under Windows, the default installation root has changed from
C:\Panoptic
toC:\Funnelback
. While theC:\Panoptic
location remains supported, we encourage you to move your installation toC:\Funnelback
if possible for consistency.