Previous Release Notes : V6R2014.SP1
 
V6R2014.SP1
 
Improvements
Upgrade notes
Major fixed issues
V6R2014.SP1 is a maintenance release to V6R2014GA. For information about new features, see the V6R2014GA Release Notes.
Improvements
Core
• Value fields and date time fields can now be used as join keys on INNER JOIN #15882.
• Control disk space use: configure the maximum number of parallel full compacts in the index builder #14263.
• Reduce index latency: new policies for dictionary build and compact allow you to schedule them independently of index build and compact #18692.
• Process documents based on their target data model class, thanks to the "Data model class" condition in the analysis pipeline #19211.
• Analysis now logs the processing conditions that led to a discarded document #19177.
• Search server can better withstand avalanches by rejecting queries above a configurable threshold
• Summary operator now adds ellipsis (...) at the beginning or end of text segment when needed #19383.
• Storing alert occurrences is now optional #19386.
• Optimized implementation of alphanum sort #19379.
• Retrieve freshly indexed documents, even with timeouts, with option to assign DIDs in reverse order #19429
• XPath extractor now supports string functions (such as substring) in addition to node sets #16312.
Semantics
• Visual editor for Fast Rules Matcher and Semantic Extractor resources in the Business Console
• CSV import and export for many resource types in the Business Console.
• Option to test a semantic resource with an existing analysis pipeline #18018.
• As a follow up to V6R2014 semantic API cleanup, you can now create custom java semantic processor and tokenizers in the semantic pipeline. See examples in sdk/java-customcode/samples. #16533.
Mashup
• Dedicated "Spellcheck List" widget for displaying "Did you mean?" suggestions, with optional number of matches #17424.
• CloudView feeds (both V5 and V6) can now connect to an access-api that's behind a proxy #14736.
• MEL functions can now be packaged and deployed as a cvplugin #18169.
• The Parameter Adapter Widget Trigger can now be triggered by widgets inside a composite widget #19148.
Connectors
• New tools to create migrators for connectors. See "Maintaining a Connector Configuration across Versions" in the Connector Programmer Guide and "Upgrading a Connector" in the Administration Guide #19374.
• Beta Sharepoint connector for Sharepoint 2010 and 2013 (requires separate install).
APIs
• Improved latency of apply changes call in a multi-host, multi-role deployment
• New functions to control the global dictionary rebuild in MAMI #10204.
• Added search API parameters to configure hit meta processing operations #14979.
Administration Console
• Use the "Browse" button to define the paths to crawl in the File System connector
• Rename Semantic Types
More user assistance
• More user-friendly CloudView launch page, with direct links to documentation #17486.
• Manipulate semantic annotations before indexing, using the SemanticPipeDocumentProcessor. See "Manipulate Semantic Annotations" in the Technical Reference. #16517.
• Practice Mashup Expression Language in the MEL playground:  http://mel.exalead.com #17583.
• Most XML editors auto-complete code, based on their schema #19039.
Upgrade notes
Migration notes
• A MaxSizeFullCompactPolicy is added to full compaction policies during migration. This ensure slots will be compacted at some point, since reindexing the same data over and over previously could ultimately trigger filesystem full #17067.
• Suggest dispatcher now sorts suggestions first by edit distance, then by frequency instead of just frequency #19618.
• Migration from versions older than V6R2013x will convert some files and may take some time (up to a few hours)
• Converted files will be renamed as .2013x.bak and may be manually removed once the migration is finished.
Migrating document cache from V6R2013x (and older)
After migration, the document cache is moved to <DATADIR>/cache.2013x.bak but is NOT migrated to V6R2014. To re-import legacy cache:
1. Start CloudView
2. Run  <DATADIR>/bin/migrate2013xCache  <DATADIR>/cache.2013x.bak/<BG BG_PAPI_HOST> <BG_PAPI_PORT>
Migrating custom Mashup applications
To migrate custom Mashup applications, you must add two properties in their development.properties file.
1. Go to custom application directory: <ROOTDIR>/sdk/java-mashupui/project/war/WEB-INF/config
2. Edit the development.properties file to add the two following properties:
o access-api.ac0.applicationId =<APPLICATION NAME> where <APPLICATION NAME> is the name of your Mashup application.
o access-api.ac0.URL = http://<HOSTNAME>:<BASEPORT+10>/access where the URL must be the same as the one specified for the  360-mashup-ui.mu0.accessapi.url  property.
3. To regenerate the WAR file, see section "Generating WAR Files for Heavy Customization" in Mashup Programmer Guide.
Changes that can break things
As of V6R2014:
• All MAMI indexing methods have changed to support new indexing server architecture. You might need to update custom developments based on MAMI.
• Previously, partial document updates with the PAPI "updateDocument" command could corrupt the index, so it's now rejected in V6R2014.
• Clearing a build group no longer clears detached replicas #19300.
As of V6R2014 SP1:
• Dictionary builder now resides on the indexing server instead of the gateway. However, for now the "Dictionary builder" role must remain on the same host as the "Gateway" role #19651.
• In Content Recommender, when "specific hit selection" is left empty, the default query is now #all  #19527.
• Certain signatures for JSP tags have changed. They use the more generic AbstractCategory instead of MergedCategory. This allows you to use those tags with the output of more generic enumerators, such as data:categoryList. Note it might break code if you were using concrete MergedCategory instead of base class #19212. Impacted tags:
o config:getOption
o render:categoryLink
o search:forEachCategoryPath
o search:getCategory
o search:getCategory2DUrl
o string:eval
o url:refine
• Crawler and FeedFetcher connectors now 
o use less disk space, by storing only extracted documents text, not the whole binary part. This means, however, that preview and download from the mashup-ui can no longer use the storage as cache.
o consume about 800MB less memory each. Reduced buffer sizes will produce a larger number of smaller files and trigger more frequent compacts.
Major fixed issues
 
Ticket
Summary
#16477
Mashup UI and API fail when Search API has basic authentication
#19060
MEL ${session & ${cookies do not work !
#18512
Using tab widget on IE8 lead to a JS Error
#18888
Properties / Mapping not working well with multivalued metas
#19238
Missing tooltips for User Query Language (UQL) and Query Expansion
#19275
Business console/Ontology edition : in test tab lemmatizer, phonetizer are disapearing
#19630
Invalid CSH links in Business Console
#19452
V6R2014 Build document cache data structures are never compacted
#19016
Lotus Notes access-restricted fields are processed but data are not extracted
#19411
Imap connector : impossible to get recipient mail address
#19183
Connector: problem with encrypted field in config
#19266
V6R2014 indexing-server does not delete old generations of dictionaries
#19162
Feed Fetcher connector is not documented
#19152
Unable to initialize connectors after switching logs path on win64
#19307
Default install script does not allow to install product correctly on aix
#19437
V6R2014 Queries may not respond when an apply changes call is performed while a dictionary generation is being built
#19569
Hotfixes fail to be uninstalled on windows platform
#16189
Crash of search-server on dynamic date facets when refined using "exclude"
#18795
Clearing buildgroup also clears detached replicas of this buildgroup
#19057
V6R2014 Incremental update of index consumes more and more memory
#19300
Clear buildgroup clears detached replica
#18683
Search reporting (csv / reporting_store) does not use retention configuration (maximum size, rotations...) for cleanup, which can lead to lot of disk space used
#11898
Clearing a frozen build group should not be allowed (now displays an error message)
#19344
ContentCleanup document processor with option stripHtml enabled can lead to crash of indexingserver process on some html files
#19481
V6R2014 Missing normalization on data mapped to value fields can lead to crash of indexingserver process
#19632
Searchable fields with no word position stored could lose the first indexed documents
#19166
Migration to V6R2014 fails if there is a "customized" related term semantic processor in the configuration
#19171
Migration to V6R2014 fails for unescaped dih entries generated by legacy connectors
#18681
Disk stats in Perf Monitoring KO with some linux filesystem mounting
#18607
Queries performed after a full compact on a non searchable category field can crash the searchserver
#18909
Adding operations on registered resource at each pipe build can crash ss0 in multi-threading context
#19200
Numerical prefix handler with useRetrieved="true" and custom key leads to replica crash
#19205
#num does not support multiple custom ranking keys : syntax error, unexpected Comma, expecting Right brace
#19319
Query "2012/10/10 00:00:00" on a datetime field is generating a range query on the full day
#19328
Summarizer : max number of segments set to 0 can lead to crash of searchserver process
#19406
Malformed or timeouted queries not reported in search.csv
#19447
Proximity boost is applied even if the words are separated by more than maxDistance
#19697
To control cvdiag file size, the search reporting data store is no longer exported