Two new appliance updates are now available: texisScripts-20.0.0 and texis-7.06.15156... The texisScripts update contains the following new features:
* For Parametric Search appliances only: The Rank Bias setting was added. This allows search results to be biased up or down in relevance-ranked searches -- on a custom, document-by-document basis -- by adding in a selected Parametric field. For example, PDF results could be ranked a few percent lower. Or documents meta-tagged `Important' could be ranked higher.
* Support for walking Azure Blob Storage was added.
* A new setting, Body Storage Method, was added which potentially speeds up some searches if Abstract Style is set to `Description'.
* The Audit Log setting was added, which allows logging setting changes (both per-profile and system wide) for auditing purposes.
* The Robots Crawl Delay Min/Max settings were added to modify robots.txt Crawl-Delay values. This allows a site with an unreasonable Crawl Delay to still be walked, while still obeying the rest of robots.txt.
* Network filesystems are now checked and remounted automatically if needed before walks start, to help avoid walk errors due to filesystem problems
Other minor features and fixes in texisScripts-20.0.0 include:
* Allow redirecting of syslog to an external receiver. (also requires thunderstonePatch newer than 2.14)
* Text-type Additional Fields can now be added to Compound Index Fields
* Log Viewer now faster, and does not repeat some dirs
* When search Phishing Protection is triggered, the errant URL is no longer displayed in the page
* Fixed Cleanup Unused Directories not being offered when needed in Disk Space Viewer
* Fixed Quotes for Literal not showing its default value of 'N' in Search Settings
* Improved search efficiency when generating abstracts from large documents
* Enhanced detection and handling of select box settings set to an invalid value
* Errors when generating CSRs now display more details about the problem
* Generate CSR form now prevents entering Country Name longer than 2 letters
* Added a Check All checkbox to Manual Installation of Thunderstone Updates
* XML sitemap files with stylesheets were being interpreted as vanilla XML (with stylesheet applied), rather than XML sitemaps
* Added CSRF protection to all forms
* A long Walk Delay -- or Crawl-delay from robots.txt -- could delay a walk's response to Stop or Pause signals; fixed
* <link rel=canonical> canonical URLs may have been made absolute incorrectly if the initial URL was a redirect and the canonical was relative
* Fixed Show Errors and Show Duplicates links when viewing a failed walk
* When making a copy of a profile, the source profile's type now overrides the profile type chosen on the form
* Fixed files removed from file servers not being removed from search index on refresh walks
* Added search setting Search Trace Settings, and renamed Trace Settings to Walk Trace Settings; allows tech-support-requested trace settings
* Debug SQL and Debug Metamorph subsumed into Search Trace Settings
* Check for running walks etc. at Update Software and warn admin; automatic updates will refuse to apply if conflict detected, check updateSoftware.log
* Fixed Exclude Extension on URLs that have query strings
* Removed unnecessary All Extensions setting
* Fixed empty result URLs when disabling PDF Highlighting and setting File URL Format to Internet Explorer Only
* Added getDocumentUsageOverview to admin SOAP API
* Increased default <XmlOutputVersion> value to 1.14: <RankBias> element added if Rank Bias used
The texis-7.06.15156... update contains the following fixes. It is also required by the texisScripts-20.0.0 update:
* Fix possible "ABEND signal 11" in aspell spell-checker if no alphanumeric words in walk data
* Fix relevance-ranked text search that requires post-processing might have failed to return the highest ranked row(s), especially if their rank(s) without post-processed terms would be lowest
* Fixes to avoid potentially truncating Rank Bias relevance-ranked results
* Relative (non-absolute) URLs in <base href> were being ignored, per earlier HTML specs
* Fix for when a PDF with a font with a character code of 0xffffffff could cause an ABEND
* Fix for when some PDF files may cause an ABEND during processing if possibly corrupt
These updates are available via the System -> System Setup -> Update Software menu.
* For Parametric Search appliances only: The Rank Bias setting was added. This allows search results to be biased up or down in relevance-ranked searches -- on a custom, document-by-document basis -- by adding in a selected Parametric field. For example, PDF results could be ranked a few percent lower. Or documents meta-tagged `Important' could be ranked higher.
* Support for walking Azure Blob Storage was added.
* A new setting, Body Storage Method, was added which potentially speeds up some searches if Abstract Style is set to `Description'.
* The Audit Log setting was added, which allows logging setting changes (both per-profile and system wide) for auditing purposes.
* The Robots Crawl Delay Min/Max settings were added to modify robots.txt Crawl-Delay values. This allows a site with an unreasonable Crawl Delay to still be walked, while still obeying the rest of robots.txt.
* Network filesystems are now checked and remounted automatically if needed before walks start, to help avoid walk errors due to filesystem problems
Other minor features and fixes in texisScripts-20.0.0 include:
* Allow redirecting of syslog to an external receiver. (also requires thunderstonePatch newer than 2.14)
* Text-type Additional Fields can now be added to Compound Index Fields
* Log Viewer now faster, and does not repeat some dirs
* When search Phishing Protection is triggered, the errant URL is no longer displayed in the page
* Fixed Cleanup Unused Directories not being offered when needed in Disk Space Viewer
* Fixed Quotes for Literal not showing its default value of 'N' in Search Settings
* Improved search efficiency when generating abstracts from large documents
* Enhanced detection and handling of select box settings set to an invalid value
* Errors when generating CSRs now display more details about the problem
* Generate CSR form now prevents entering Country Name longer than 2 letters
* Added a Check All checkbox to Manual Installation of Thunderstone Updates
* XML sitemap files with stylesheets were being interpreted as vanilla XML (with stylesheet applied), rather than XML sitemaps
* Added CSRF protection to all forms
* A long Walk Delay -- or Crawl-delay from robots.txt -- could delay a walk's response to Stop or Pause signals; fixed
* <link rel=canonical> canonical URLs may have been made absolute incorrectly if the initial URL was a redirect and the canonical was relative
* Fixed Show Errors and Show Duplicates links when viewing a failed walk
* When making a copy of a profile, the source profile's type now overrides the profile type chosen on the form
* Fixed files removed from file servers not being removed from search index on refresh walks
* Added search setting Search Trace Settings, and renamed Trace Settings to Walk Trace Settings; allows tech-support-requested trace settings
* Debug SQL and Debug Metamorph subsumed into Search Trace Settings
* Check for running walks etc. at Update Software and warn admin; automatic updates will refuse to apply if conflict detected, check updateSoftware.log
* Fixed Exclude Extension on URLs that have query strings
* Removed unnecessary All Extensions setting
* Fixed empty result URLs when disabling PDF Highlighting and setting File URL Format to Internet Explorer Only
* Added getDocumentUsageOverview to admin SOAP API
* Increased default <XmlOutputVersion> value to 1.14: <RankBias> element added if Rank Bias used
The texis-7.06.15156... update contains the following fixes. It is also required by the texisScripts-20.0.0 update:
* Fix possible "ABEND signal 11" in aspell spell-checker if no alphanumeric words in walk data
* Fix relevance-ranked text search that requires post-processing might have failed to return the highest ranked row(s), especially if their rank(s) without post-processed terms would be lowest
* Fixes to avoid potentially truncating Rank Bias relevance-ranked results
* Relative (non-absolute) URLs in <base href> were being ignored, per earlier HTML specs
* Fix for when a PDF with a font with a character code of 0xffffffff could cause an ABEND
* Fix for when some PDF files may cause an ABEND during processing if possibly corrupt
These updates are available via the System -> System Setup -> Update Software menu.