TOLERANT Match Release 7.0

Notices

• The Match project attribute autoLoad is deprecated now and will be removed in future releases.
• The new trimtypes M, RM, LM and A have been added to the following configuration-elements and tools:
  • serviceInputField
  • serviceOutputField
  • csvInputField
  • csvOutputField
  • databaseInputField
  • databaseOutputField
  • matchFiletest.sh/bat
  • postFiletest.sh/bat
• Limitation: When a paging database and an original data database are configured simultaneously, the number of allowed full savepoints is limited to one.
• The installer will abort installation, if the target directory is not empty. The following exceptions are allowed: the installer itself can be in the installation directory or a previous installation can be in the installation directory.
• The installer now always queries the local hostname, even if HTTPS is not used. This name is stored as environment variable in the „tolerantEnv“ script, in order for the Admin CLI to use the correct hostname instead of „localhost“. The silent installer properties now support this new field as „tlhostname“. If not set, „localhost“ will be used, as before.
• All logfiles other than the request log now have a new timestamp format. It now complies with the ISO 8601 standard. Examples:
  Previous: 20 Mar 2017 16:27:58,840
  Now: 2017-03-28 16:27:58,840
• An up-to-date java virtual machine (1.8.0_131) is now bundled with the product.
• The default memory allocator on Windows has changed. Impact: About 25% decrease of occupied memory / About 5% increase of runtime.
• Example configurations, templates and API examples have been overworked.

New Product Features

Configuration and Administration GUI

• A template for custom transliterations was missing. If a custom transliteration is created through the admin GUI, a template will now be used to create a basic transliteration rule file.
• More rule field attributes are displayed in GUI by default
• Connection check in admin GUI now works for a H2 database that is currently in use by service.
• The execution dialog for initial load now shows the input file as well as the rejected files.
• Several improvements were made for the administrative GUIs
• New menu item in the context menu of the Admin-GUI to generate an identical config-file to the running service
• Using Match Administration GUI, you can now configure the matchProfileField attribute extDiffScore

Client GUI

Added new „info“ menu entry in client GUI

Batch

The match anonymizer tool did not report lines with identical keys. This lead to problems, as the result is no longer uniquely identifiable. Now a list of non-uniques will be reported at the end.

Service

• The new attribute „forceOutput“ for ServiceOutputFields has been added. It can be used to force the ServiceOutputFields to appear in a service response even if the field is not mapped with an InputFieldmapItem. In this case, its value is an empty string.
• When a paging database and an original data database are configured simultaneously, the number of allowed full savepoints is limited to one.
• A new match engine of type PROXY can now be used to send requests from a local Match service to a remote Match service using pseudo anonymization.
• Some REST client frameworks cannot create PUT/DELETE requests. In such cases the X-HTTP-Method-Override for POST requests can be used.
• Cyclic connection and performance test of communication in Match cluster
• Extended performance logging for database access: effective bandwidth and latency
• Extended logging of performance metrics during cluster synchronization
• Monitoring of divergent clocks of cluster nodes. If difference is too large, cluster nodes will not start now.
• Improved detection of cluster sync partner if system clocks of cluster nodes differ more than approx. 200ms
• Soap response of a delete operation returns now the service error message, if the key could not be deleted

General

• New trimtypes have been added to input and output fields
• emptyScore is now supported for partitioning fields. Please note that configuring emptyScore for partitiong field could lead to reduced performance.
• Extended logging of environment information
• Support information extended by last modification date for each file
• A new Match field of type DATE can now be used to store and query date values.
• Improved installation consistency check in ’support.sh/bat‘ and ‚checkinstallation.sh/bat‘
• Using the new matchRule attribute maxResultPercentage you can now weight all results of a certain rule, so that results found using a fallback rule would have a lower score value.
• Logging of system and process memory footprint if out of memory condition occurs in Match kernel
• An up-to-date java virtual machine (1.8.0_131) is now bundled with the product.
• The default memory allocator on Windows has changed.
• New support function to determine maximum available memory (option -x of matchKernelUtility)
• Overworked several configurations and API examples that come with the product
  • C-API
  • Java-API
  • PL/SQL-API
  • Batch examples
  • Service examples
  • Service templates

Security

Several improvements were made for the administrative GUIs

API

PL/SQL client can now handle SSL-connections using the new method createSSLConnection

Fixed Bugs

Configuration and Administration GUI

• TPR-3901: Path validation for normalizer files was improved
• TPR-3977: Trying to edit a synonym file in the admin GUI when no filepath was set, led to internal GUI errors. It is now no longer possible to edit a synonym file, if the path is not set.
• TPR-4389: Synonyms now cannot be added to fields if no synonym file has been configured
• TPR-4436: Renaming newly added profiles will now take effect on focus change

Client GUI

• TPR-3881: Consistent display of stopped projects in client GUI
• TPR-4440: If the first project in a configuration was not running correctly, the client GUI did display the input fields of the (not running) first project upon startup, but switched internally in the project dropdown-selector to the second project. Subsequent requests might thus have failed, if input field names between the two projects were different. This behavior is fixed now.

Batch

• TPR-4409: The output of durations in batch processing did use different locales and was either using „,“ (comma) or „.“ (dot) as decimal separator. This has now been fixed.
• TPR-4419: The metric names for synonyms and normalizer metrics were corrupted after the client GUI was started. The names are now no longer affected by the start of the client GUI.
• TPR-4433: Match Batch ensures that no lines from the self file are contained in clean file, even if the input file had some faulty lines.

Service

• TPR-992: The Match project attribute autoLoad is deprecated now and will be removed in future releases.
• TPR-1566: Match projects that are in ERROR state will be ignored during the cluster synchronization phase
• TPR-3880: The service response for soap request with security activated was changed, so that the service’s error code is written in soap body
• TPR-3951: Error occurred when acquiring internal data (resultMode=4) for a matchProject with STOREONLY fields.
• TPR-3953: Match admin command „save project“ did not work if the project was started as a cluster project
• TPR-4004: Fixed SOAP delete response to match WSDL
• TPR-4050: Cluster Synchronization used old (historic) backlog files where it should not
• TPR-4051: Full cluster synchronization failed if multiple deletes for same key were contained in backlog file
• TPR-4064: Match Initialload now stops after a match kernel error was encountered
• TPR-4065: Unnecessary large kernel table may have caused OutOfMemory during Initial Load
• TPR-4070: The retry mechanism for acquiring lock now considers the possibility of having an existing lock entry from the same service.
• TPR-4071: Table prefix is not considered when checking for database uniqueness
• TPR-4072: The installer now always queries the local hostname, even if HTTPS is not used. This name is stored as environment variable in the „tolerantEnv“ script, in order for the Admin CLI to use the correct hostname instead of „localhost“.
• TPR-4137: Wildcard delete operations from original database will be carried out only if the delete operation was successful in Match kernel.
• TPR-4198: In case of windows operation system, variable MY_MEM_OPTS could now be overridden by service registration.
• TPR-4340: Fixed ‚matchInspectPagingDatabase.sh/bat‘ for SQLServer database
• TPR-4402: If both HTTP and HTTPS are configured the admin CLI will now prefer HTTPS.
• TPR-4423: Match Initialloader returns now a generic error code if the entered project was not found
• TPR-4447: Internal data mode no longer returns „null“ values in service responses.

General

• TPR-3519: Improved misleading installer output for modified files
• TPR-3550: The Support script now uses the temporary folder of the installation
• TPR-3839: The installer will now create a backup version of the old tolerantEnv file
• TPR-3840: Upgrade now deletes the Jetty-tmp directories if the hostname was used for JETTY_HOST
• TPR-3843: Installer-Script now resolves symbolic links
• TPR-3851: SaveOnShutdown will be skipped if Match was installed without service
• TPR-3892: The new trimtypes M, RM, LM and A now work for the following configuration-elements and tools:
  • serviceOutputField
  • csvOutputField
  • databaseOutputField
  • matchFiletest.sh/bat
  • postFiletest.sh/bat
• TPR-3937: The installer will abort installation, if the target directory is not empty.
• TPR-4000: Calculation of database statistics were improved
• TPR-4065: Unnecessary large kernel table may have caused OutOfMemory during Initial Load
• TPR-4069: Improved tracing in case of kernel errors
• TPR-4083: All logfiles other than the request log now have a new timestamp format. It now complies with the ISO 8601 standard. Examples:
  Previous: 20 Mar 2017 16:27:58,840
  Now: 2017-03-28 16:27:58,840
• TPR-4165: Initial load crashed for very long multi string fields with many common tokens
• TPR-4183: Calculating the total memory foot print was corrected for Windows systems. Note that the memory foot print will be calculated only if the tracing level was set to DEBUG or higher.
• TPR-4376: The script that verifies an installation did report some changed files multiple times. This has now been fixed.
• TPR-4381: The gui.properties file had windows line-endings even on *nix platforms. This has been fixed.

Security

TPR-3878: Improved error messages for ’security.bat/sh‘ in case of wrong usage

API

TPR-4442: PL/SQL API now returns an error message, if needed internal fields are missing from the configuration.

Known Bugs

Currently no known bugs.