*1.1.1 (March 27th, 2009)* * Support for the option for subscribing service automatically to the BLACKBOARD/LAST_REQUEST element, in order to receive messages for the orchestration protocol *1.1.2 (Mai 7th, 2009)* * The new Yadda version integrated. Wojtek's notes about important Yadda changes: - new management of filters (BBQ); we decided to make filter management a bit more explicit; in 0.7.1 SearchQuery could contain FilterCriterion (search criterion + filter name) which resulted in creation of filter matching documents found with specified criterion; now filters are created with FilterDefinition and SearchQuery may be passed with filter name; there are at least two advantages of new approach: -- user may explicitly delete or recreate existing filters -- user may create not only simple filters but also boolean filters which combine filters created earlier old API still exists but is deprecated; I did not change old approach in driver and so driver uses deprecated filter API now one more remark about filters - our tiki documentation describes services2 API for search which is not currently used in driver facade (maybe one day driver facade could be switched to services2 API which seems cleaner than using SearchModule directly); filter management is still possible - just use FilterFactory returned by SearchModule.getFilterFactory() method; FilterFactory has addFilterDefinition() and removeFilterDefinition() methods which manage filter definitions: /** * Adds filter definition. If filter definition with matching name already exists it is replaced if and * only if replace parameter is true. */ public void addFilterDefinition(FilterDefinition filterDefinition, boolean replace); /** * Removes filter definition with specified name. Note that boolean definitions containing filters with * matching name will be removed recursively. */ public void removeFilterDefinition(String filterName); - new management of analyzers: this change is transparent to driver's java code - I already did necessary changes in driver facade (in my branch); what is not transparent is change of index definitions in spring - that's why old indexes must be dropped with 0.7.3 update; few words about this change: analyzers in index configuration are not defined by analyzer's java class anymore; AnalyzerFactory in SearchModule contains mapping between simple names and analyzer's classes; default analyzer for index and analyzers for single fields are defined just with their names *1.1.3 (Mai 18th, 2009)* * Service Common Configuration adopted; support for Perl services removed *1.1.4 (Mai 19th, 2009)* * The dependencies in ivy.xml changed from last.integration to 0.0.+ *1.1.5 (June 4th, 2009)* * wrong service endpoint reference in properties fixed (#646) SWITCHING ON BUILD SYSTEM VERSIONING *0.0.1 (June 8th, 2009)* * version changed, reduced from 1.1.5 to 0.0.1 * proper EPR property for service registration set in the context file * ResultSetIterator consider now the status of resultset, it waits for the records until RS is closed * context and endpoint properties modified *0.0.2 (June 27th, 2009)* * Bug #689 fixed: Index Service uses now CNR DateUtils class for setting proper date and timezoen in DS profiles *0.0.3 (July 15th, 2009)* * reversing patch r2984 (#550) *0.1.0 (October 8th, 2009)* * some of the classes and interfaces exported to icm-commons * using profiles and blackboard classes from icm-commons * support for full text indexing: - download service for downloading full texts integrated - new message parameter 'merge' intoduced, if the parameter is true then update on lucene index record is processed * API changed: new methods createIndex, feedIndex, deleteIndex added (#824) *0.1.1 (October 13th, 2009)* * the release 0.1.0 was broken because of wrong dependencies in icm-commons (icm-commons pointed on the older icm-functionality-api) *0.1.2 (November 18th, 2009)* * introduced new XML based MDFormatRelationsContainer for storing information about the mdformats and layouts in XML file representation (JAXB) - ticket #745 file representation * merging meta-data and full-text fields in Lucene introduced * API changed: boolean merge parameter introduced for feedIndex method *0.1.3 (December , 2009)* * removing deleted records from Lucene index introduced (#337) * modification for proper errors handling *0.1.4 (January 25th, 2010)* * the new Yadda packages integrated * new method deleteDocumentByTerm(String indexId, String field, String value) introduced: Methods deletes from the index all documents containing specified value in a field. Field must be searchable and untokenized - note that these conditions mean that value is case sensitive *0.1.5 (January 25th, 2010)* * new method undeleteAll(String indexId) introduced: Undeletes all documents marked for deletion. Note that documents marked for deletion are documents deleted with one of {@link #delete(Collection)}, {@link #delete(String, String)} methods since last index optimization. During optimization all documents marked for deletion are physically removed from the index. *0.1.6 (September 23th, 2010)* * the feeding process is split now on chunks where each chunk contains the number of records specified by services.index.feeding.chunk.buffer.limit property; For each chunk process opens index session, runs chunk content feeding and then commits the session. * asynchronous downloading of full-texts * API changes: new fulltext specific parameter added to the feedIndex method as well as to the notification message; parameter specifies if the feeding is the fulltext feeding or standard metadata feeding. *0.1.7777777 (, 2010)*