====================================================================== Endeca MDEX Engine GENERAL (6.2.2) * Release Date: March 2012 * The release notes may have been updated since the release date. Contact Oracle Endeca Customer Support for the most recent version of the release notes. BUG FIXES (6.2.2) Dgraph (6.2.2) * IFCORE-1414: The Stratify relevance ranking module now allows underscores in its name. * IFCORE-1527: Dgdix prints out a warning if the language code is missing in the stemming-updates filename. * IFCORE-1078: Derived properties no longer round floating point values for the MIN and MAX functions. * IFCORE-1547: Queries containing the sequence ":::" or "___" could cause the Dgraph to return corrupted results which could cause the APIs to crash. * IFCORE-1553: The Dgraph would sometimes return incorrect counts from dimension value search when the query included a Dn parameter with selections from a multi-select OR dimension. * IFCORE-1560: The Dgraph would fail when serving text search queries after a config update of an application's thesaurus entries. * IFCORE-1561: The Dgraph would fail when processing a compressed update file. * IFCORE-1565: The Dgraph could fail when using a relevance ranking strategy. Request Log Analyzer (6.2.2) * Standard Deviation: There was a bug in the way Cheetah computed standard deviation of reported statistics. The new figures are correct. * Hourly Reports: The hour range of hourly reports may differ by one hour from Cheetah. The statistics from the hourly report will not change. * Worst Query List: Cheetah was incorrectly ignoring some requests for the worst query list. You may now see queries that were not reported by Cheetah. ====================================================================== Endeca MDEX Engine GENERAL (6.2.1) * Release Date: November 2011 * The release notes may have been updated since the release date. Access the Product Downloads section of the Endeca Developer Network (https://eden.endeca.com) for the most recent version of the release notes. MIGRATION (6.2.1) * The silent installer on UNIX no longer prompts for acceptance of the License Agreement. In the text file that you create for the silent installation, remove the line that contains "Y" to accept the agreement. BUG FIXES (6.2.1) Dgraph (6.2.1) * BUG19698: Stratify relevance rank modules sent in via the Nrr parameter which contain spaces would return incorrect results. * BUG19704: In some case Geocode filters would return no results. * Bug Fix (IFCORE-1492): When combined with text search filters, in some cases hierarchical multi-or dimensions would return dead ends as valid refinements. * Bug Fix (IFCORE-1497): An EQL expression which OR'd the results of AND expressions did not return correct results. * Bug Fix (IFCORE-1511): If the spell+nospell option was turned on certain search terms would return incorrect results. * Bug Fix (IFCORE-1521): Dimension ids greater than 2147483647 would not be parsed properly by the XQuery API. * Bug Fix (IFCORE-1532): Range filtering on a property with a high degree of multi-assign in combination with a record filter could cause the range filter to be ignored. KNOWN ISSUES (6.2.1) Documentation (6.2.1) * The documentation for this release has not been updated. Updated documentation may be provided in a future release if necessary. Please see the EDeN knowledge base for the most recent documentation. * REM-50: You no longer need to download and configure separate license keys to enable Chinese, Japanese, or Korean languages. Support for these languages is enabled by default as part of MDEX Engine installation. (You can disregard "Configuring license keys for language support" in the MDEX Installation Guide.) ====================================================================== Endeca MDEX Engine GENERAL (6.2.0) * Release Date: July 2011 * The release notes may have been updated since the release date. Access the Product Downloads section of the Endeca Developer Network (https://eden.endeca.com) for the most recent version of the release notes. BUG FIXES (6.2.0) Dgraph (6.2.0) * CHG13086: The new "Why Rank" feature may be used to report information about how relevance ranking is used when ordering records. It is enabled on a per-query basis and is usable both with record search and dimension search. For more information, see the MDEX Engine Advanced Development Guide, in the "Using Why Rank" section of the "Understanding and Debugging Query Results" chapter. * CHG03924: The Why Match feature is now enabled per query and returns the fields and query terms that matched the record returned. The --whymatch and --whymatchConcise flags are deprecated. * BUG08380: Enabling Invalid/Empty Dimensions for record sort no longer causes the ERecSortKeyList to not be populated. * BUG19071: The performance of aggregated record refinement counts (using --stat-abins) was improved. * BUG16153: In some cases, the warning "Langauge (no language specified) not specified in stemming file" would be produced. In such cases, the language is now correctly identifed. * BUG19417: In rare cases, incorrect results could be returned for a phrase search if a previous search for an identical phrase returned no results for a different search interface. * BUG19160: Specifying an excessive number of unique language IDs would cause the MDEX Engine to eventually fail. * CHG18788: Error reporting has been enhanced when range filtering on an invalid property type. * BUG19026/BUG18211: Duration and Datetime types could be incorrectly included or excluded from range filters. * CHG04012: Partial updates did not accept fully-featured DVAL or DIMENSION_ADD_SUBTREE elements within a DIMENSION_VALUE_ADD. * CHG17855: The /config?op=update command does not pick up changes to merchandising groups, zones, and styles. * CHG12895: Allow dimension search against a set of dimensions. * CHG13346: Add feature to disable a dimension from dimension search. * CHG13265/CHG13898: Specify a collation preference (e.g. "standard") with the argument --lang to enable one of the standard sort orders available for words with international characters. * CHG13252: Refinement counts can be returned for dimension search. This ability can be turned on or off on a per-query, per-dimension basis. Refinement counts have also been given this level of control. * BUG18326: Standalone relrank was ignored if relrank terms matched no results. * BUG18598: Fixed the alpha sort order of properties and dimensions to sort without regard to case and to use case only to break ties. Lower case now sorts before upper case. * BUG18855: Fixed a MDEX Engine failure when a property in record_sort_config.xml started with "Endeca." (for example, "Endeca.Identifier"). * BUG18813: Fixed a bulk export bug that occured when the number of properties, dimensions, and records was larger than approximately 4 billion. * BUG18801: Fixed a bug in scoring and filtering DYM suggestions from the word break analysis. * BUG18874: Fixed multiple memory corruption issues that occured in rare cases. (Empty GET requests, or GETs for files with very short names could sometimes cause the MDEX Engine to fail.) * BUG19072: Disk contention between MDEX logging and the load from partial updates and background merges can cause sporadic increases in query processing latencies. * Deprecation: The WEIGHTED_FREQUENCY option for Dynamic Ranking of dimension values has been deprecated. This behavior has been superseded by dimension value boost. Agraph (6.2.0) * BUG18852/BUG14138: Fixed a bug in summing refinement counts from Dgraphs which produced incorrect ordering of the refinements. Dgidx (6.2.0) * CHG19311: Indexing time has improved in the case where there are thousands of assignments or more on records. * CHG18599: Allow the use of a custom stemming dictionary. * CHG18486: Updated the memory allocation scheme to more effectively deal with many properties and dimensions. Dgwordlist (6.2.0) * BUG15900: The dgwordlist --version flag returned an incorrect version number. The version number now tracks the Dgraph and Dgidx. DOCUMENTATION (6.2.0) * BUG18953: The Basic Development Guide containted Java examples of stratification API methods that were incorrect. * Bug Fix (BUG18606): The Web Services and XQuery Developer's Guide contained an incorrect property name for DGraph.IsDisabledRefinement in text and code samples. The correct name of the property is DGraph.DisabledRefinement. * BUG Fix (BUG18607): The Web Services and XQuery Developer's Guide was missing the DisabledRefinementsConfig element from the RefinementConfigList data type example. * Bug Fix (BUG18597): The Web Services and XQuery Developer's Guide was missing the Sort element from the AggregateRecordDetailsQuery data type example. * Bug Fix (BUG18595): The Basic Development Guide contained an incorrect property name for DGraph.IsDisabledRefinement in text and code samples. The correct name of the property is DGraph.DisabledRefinement. The section on "Identifying disabled refinements from query output" has been corrected. ======================================================================