*0.0.2 (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 *0.0.3 (Mai 19th, 2009)* * Global common configuration of properties adopted * No more support for Perl services *0.0.4 (June 4th, 2009)* *no proper service end point reference fixed *contex and epr properties modified *0.0.5 (June 27th, 2009)* *using CNR DateUtils library for handling DS profile dates *0.0.6 (June 28th, 2009)* *Consideration of open/close state of resultset by storing records *Blackboard profile and other profiles utilities moved in to the unibi-commons *0.0.7 (September 25th, 2009)* * bug #787 - not correct order override properties * xml XPathParser moved to unibi-commons * blackboard support moved to the icm-commons * API changed: input parameter 'rsEPR' of method storeMDRecordsFromRS changed from String to W3CEndpointReference *0.0.8 (October 8th, 2009)* * API changed, method deliverMDRecords: - parameter 'mdformat' removed - new parameter 'sortByIdentifer' for sorting results by objectIdentifier introduced *0.1.0 (November 30th, 2009)* * bug #901 for DELETE notification fixed: checking mdstoreID parameter for empty strings, missing BB Failure notifications added. * bug in generating record paths, code cleaned * the algorithm for creating mdstore directories tree has been changed; the mdstores of different formats are not any more stored in different directories; the mdstore format can be identified in the file system through the empty file "format-" stored inside the mdstore directory; the mdstore directory path is build now in the following manner: (1) basicMDStoreDirectory(taken from properties (2) hash code generated from mdstore identifier (limited to the 3 first numbers - maximum) (3) mdstore identifier it means: /(1)/(2)/(3) * the algorithm for creating record paths has NOT been changed: (4) record identifier length (5) hash code generated form record identifier (limited to the 4 first numbers - maximum) (6) record identifier (limited to 255 characters) it means: /(1)/(2)/(3)/(4)/(5)/(6) * Content of mdstore Lucene index changed: - stores 'X' character instead of mdstore id (its enough for searching - returning all records from index) - stores only record hash path instead of absolute record path; the complete records paths are build now during delivering records to the client; *0.1.1 (March 10th, 2009)* * dropping exception in mergeMDRecordKeywords in case if the input field name is defined without the schema. * Lucene index was not closed properly when the exception occurred during feeding mdstore * IndexReader in LuceneStorageReader was not properly closed; now it has been removed at all and replaced directly by IndexSearcher' * locking parallel storing processes on the same mdstore data structure * cleaning data structure process blocked response to the client because it was not inside the thread, now it is. * automatic repairing Lucene indicies and removing index locs * the statistic fields in MDStore profile were not properly cleaned during REFRESHING * API changes: - the return type of storeMDRecords changed from boolean to void, additionally the MDStoreServiceException is returned - option for running storing process via notification (STORE) - new methods createMDStore and deleteMDStore available via IMDStoreServiceDef interface * migration class for transforming mdstore data structures in version 0.0.+ to the version 0.1.+ -