Senzing Release Notes

Version: 3.12.2
Released: 1 Nov 2024

1. Fixed: Issue when using multiple processes or multithreading, the ENTITY_NAME in response documents was the last cached name for that thread or process. The current best ENTITY_NAME is now shown
2. Fixed: Case-insensitive email matches have the feature behavior downgraded from F1 to FF. The feature behavior is no longer downgraded

Version: 3.12.1
Released: 23 Oct 2024

1. Improve caching of prepared statements on all databases to reduce resources and encourage replanning
2. Enhanced MSSQL error retries and logging on certain errors
3. Enhanced PostgreSQL error logging

Version: 3.12.0
Released: 19 Sep 2024

1. New: Added setting for case-insensitive email usernames in keys. Contact support@senzing.com for instructions
2. Improved scaling with multiple SQLite files
3. Fixed: Expired lock issue with very long-running unresolve calls
4. Removed PII from logging of infinite resolve messages to only include DATA_SOURCE and RECORD_ID values

Version: 3.11.2
Released: 9 Sep 2024

1. Fixed: Updating records having an ampersand symbol '&' in RECORD_ID causes an error: parser error : EntityRef: expecting ';'
2. Fixed: Under certain data conditions, reevaluate*() calls do not fully reevaluate records after a configuration change
3. Improved error handling for invalid prepared statement error messages
4. Optimized distinct address parsing

Version: 3.11.0
Released: 7 Aug 2024

1. Improved delete performance
2. Added INBOUND_FEAT_ID, INBOUND_FEAT_USAGE_TYPE, CANDIDATE_FEAT_ID, CANDIDATE_FEAT_USAGE_TYPE, SCORE_BUCKET, and SCORE_BEHAVIOR to the FEATURES_SCORES section of the G2_INCLUDE_FEATURE_SCORES flag JSON output for searchByAttributes() and searchByAttributesV3(). This standardizes the G2_INCLUDE_FEATURE_SCORES JSON output across calls

Version: 3.10.4
Released: 30 Jul 2024

1. Fixed: On PostgreSQL, an error, 1001E|Critical Database Error '(-1:another command is already in progress )', can occur when database processes are not closed properly
2. Fixed: Under certain rare data conditions, a reevaluation can cause continuous retries on ambiguous entities
3. Fixed: First character transposition bug in the Damerau–Levenshtein implementation
4. MSSQL: Improved error handling cleanup and logging to stderr

Version: 3.10.3
Released: 12 Jun 2024

1. Fixed: getRecord() returns 0039E|No observed entity for entity key during an update
2. Fixed: Issue where DOBs can under-match if some values have timestamps
3. MSSQL: Transaction Zones are now disabled by default when using a clustered database

Version: 3.10.2
Released: 23 May 2024

1. Fixed: PostgreSQL database library dependencies
2. Fixed: Potential crash when using G2Engine.destroy() with MSSQL

Version: 3.10.1
Released: 16 May 2024

1. Fixed: Issue where the Python SDK reevaluateEntityWithInfo() fails for ENTITY_ID values > 4,000,000,000
2. Fixed: Scoring issue for features with multiple qualifiers (e.g., country and type) using the ID_COMP comparison algorithm

Version: 3.10.0
Released: 6 May 2024

1. Added: New senzingrepo v2 repository packages that detect architecture and platform. The new repository packages work only for versions >= 3.10.0. If a prior Senzing version is required, you must install the older senzingrepo v1 repository packages. See https://docs.senzing.com/quickstart/quickstart_api/index.html or contact support@senzing.com for instructions
2. Added: Support for OpenSSL 3.0. NOTE: Existing Senzing licenses are for OpenSSL 1.1, if you are installing a new Senzing version that supports OpenSSL 3.0, you must contact support@senzing.com to request a new OpenSSL 3.0 compatible Senzing license. If your license is not compatible, this error is given: ERROR: senzing.G2Exception.G2LicenseException: 999E|Invalid License (Invalid Signature)
   • The new license packages support Senzing 3.10 or newer regardless of OpenSSL versions
   • If you are using a Senzing version prior to version 3.10.0, you must use the older license files
3. Enhanced performance on large entities due to repeated data. e.g., same ER data but a new value like an ID is added each time. Performance for this data situation is improved up to 3x!
4. Enhanced performance on large "furball" tests due to many varied values. Performance for this data situation is improved up to 4x!
5. Optimized database queries resulting in 30% fewer select statements and 44% fewer rows returned compared to 3.9 when running ICIJ data as a benchmark
6. Reduced redo record creation
7. A1 and A1ES features no longer go generic
8. Improve compatibility with MSODBC v18
9. Changed name of Senzing's libpostal library build allowing developers to use Senzing and the public libpostal projects (e.g., pypostal) in the same process
10. Fixed: Issue where why*() calls fail when retrieving corrupted entities
11. Fixed: Under certain rare data conditions, an orphaned record that is reevaluated creates a relationship to itself
12. Fixed: script-based name culture detection

Version: 3.9.1
Released: 17 Apr 2024

1. Optimized database reconnect logic and performance. DB2, MSSQL, and Oracle in particular improved

Version: 3.9.0
Released: 12 Mar 2024

1. Fixed: Under certain rare data conditions, when scoring multiple features, a denial calculation may be skipped causing a relationship to be deemed a resolution
2. Added the locklist to retry timeout messages
3. Improved a SQL JOIN statement for better PostgreSQL optimizer handling
4. Various fixes and improvements

Version: 3.8.3
Released: 14 Feb 2024

1. Fixed: Issue where why*() calls fail when retrieving corrupted entities
2. Added confused entity feature display on get*() calls
3. Improved database connection drop detection, retry, and recovery functionality
4. Various fixes and improvements

Version: 3.8.2
Released: 16 Jan 2024

1. Fixed: Under certain rare conditions, excess redo records were generated which could cause incomplete redos if coupled with database connectivity issues. To check if the issue is affecting your Senzing deployment, run the following SQL: select oe.OBS_ENT_ID from OBS_ENT oe LEFT JOIN RES_ENT_OKEY reo ON oe.OBS_ENT_ID=reo.OBS_ENT_ID where reo.RES_ENT_ID is null; If any rows are returned, contact support@senzing.com for instructions
2. Optimized redo record processing
3. Optimized entity reevaluation
4. Various fixes and improvements

Version: 3.8.1
Released: 21 Dec 2023

1. Improved performance of LIB_FEAT and RES_FEAT_STAT table inserts
2. Fixed: Issue where how*() and why*() calls do not throw Unknown resolved entity value exceptions for rare conditions where RES_ENT entries do not have corresponding RES_ENT_OKEY entries
3. Various fixes and improvements

Version: 3.8.0
Released: 8 Nov 2023

1. Improved performance by consolidating insert/update/upsert operations into single SQL calls to reduce network chatter and reduce XID usage (PostgreSQL)
2. Improved performance of address transliteration by storing transliterations of parsed addresses
3. Improved MM/DD swap handling for partial DATE_OF_BIRTH values
4. NAME_KEYs now generate redo records for new installations. This helps address "stale" matches on NAME+DOB and possible matches on NAME only. NOTE: This change does dramatically increase the amount of redo records generated during load. If you prefer to disable this on a new installation, contact support@senzing.com for instructions
5. Added new whyRecordInEntity() function to determine why a particular record resolved to an entity
6. Added support for TrustServerCertificate=yes in MSSQL v18 connection strings
7. Added support for ; in passwords. NOTE: Both Microsoft and ODBC 3.0 recommend avoiding these characters in connection details: ”A DSN or connection string value enclosed with braces ({}) containing any of the characters []{}(),;?*=!@ is passed intact to the driver. However, when using these characters in a keyword, the Driver Manager returns an error when working with file DSNs but passes the connection string to the driver for regular connection strings. Avoid using embedded braces in a keyword value.”. See https://learn.microsoft.com/en-us/sql/odbc/reference/syntax/sqldriverconnect-function
8. Added support for automatically leveraging deterministic and non-deterministic encryption methods depending on the engine's needs for a particular field. Encryption plugins will need to be updated with new functions
9. NEW: Pre-release Linux ARM packages and docker hub images are available, contact support@senzing.com for details
10. Fixed: Issue where WithInfo response message AFFECTED_ENTITIES lists would include non-existent virtual ENTITY_ID values
11. Fixed: Issue during certain data conditions where specific organization cultural variants were not being leveraged in matching, and instead generic or ambiguous variants were used
12. Fixed: SQL column case if MSSQL is set as case-sensitive
13. Various fixes and improvements

Version: 3.7.2
Released: 19 Oct 2023

1. Encryption plugin capability changes:
   • Added support for digital signatures (one-way hashes) to be encrypted. NOTE: If you are already using an encryption plugin, contact support@senzing.com for details
   • Added support for non-deterministic encryption algorithms for fields that allow such behavior. Plugins will need the two new functions.
   • Fixed: potential crash in plugin call
   • Ask about our example using GoLang to implement an encryption plugin!
2. Fixed: MSSQL statement column case for databases configured to be case sensitive
3. Various fixes and improvements
4. Updated license agreement

Version: 3.7.1
Released: 26 Sep 2023

1. Fixed: MSSQL issue reading NULL clob values. This issue can occur when retrieving data for entities loaded in versions < 3.0.0 that have not been touched since
2. Fixed: G2ConfigTool.py commands used to upgrade 2.x to ≥ 3.5.0 versions

Version: 3.7.0
Released: 6 Sep 2023

1. NEW: searchByAttributesV3() includes a searchProfile parameter that allows searches to use different generics thresholds. Please contact support@senzing.com for details
2. Fixed: G2Loader.py PostgreSQL governor URI parsing for encoding special characters
3. Improved detection, correction, and reporting of corrupted entities
4. Improved ER performance of entities with many name features
5. Improved performance during transliteration
6. Added support for non-odbc.ini MSSQL connection strings to better accommodate containerized deployments: CONNECTION=mssql://username:password@server:port:database/?driver=mssqldriver
7. Various fixes and improvements

Version: 3.6.2
Released: 24 Aug 2023

1. Fixed: Under certain rare name data conditions, relationships may be logged with a null MATCH_KEY value

Version: 3.6.1
Released: 8 Aug 2023

1. Fixed: Variants might not always be correctly leveraged during generic string scoring under certain data conditions. Does NOT impact name scoring
2. Fixed: Incorrect country code assigned under certain data conditions

Version: 3.6.0
Released: 28 Jun 2023

1. Optimized database communications
2. NEW: Added Burmese personal name culture support
3. Improved performance comparing Han organizational names mapped to personal names fields, NAME_FIRST, NAME_MIDDLE, NAME_LAST, or NAME_FULL (non NAME_ORG fields)
4. Improved performance comparing Chinese organizational names
5. Improved performance of searches that don't return features (the improved performance does apply to returning feature scores)
6. Enhanced address scoring. Low quality address parses are now penalized to prevent overmatching
7. NEW: Searches are now read-only. get*(), howEntityByEntityID, searchByAttributes(), why*() calls are all read-only
8. EARLY EVAUATION FEATURE: Senzing can connect and perform read-only operations against read-only databases (replicas, standby, etc...). Please contact support@senzing.com for help
9. Various fixes and improvements

Version: 3.5.3
Released: 2 Jun 2023

1. Fixed: A WithInfo issue occurring in certain data conditions where an existing record in a non-singleton entity is updated causing it to split off into a new singleton entity. The WithInfo AFFECTED_ENTITIES list only contained the new singleton entity

Version: 3.5.2
Released: 3 May 2023

1. Improved database lock creation and handling. Fix for exec called without holding lock error

Version: 3.5.1
Released: 17 Apr 2023

1. (Python tooling) Fix for G2UpdateProject.py erroring when trying to find the non-existent JRE: FileNotFoundError: [Errno 2] No such file or directory: '/opt/senzing/g2/lib/jdk-11.0.10+9-jre. If a Senzing project upgrade to 3.5.0 failed, upgrading to 3.5.1 fixes the project
2. Various fixes and improvements

Version: 3.5.0
Released: 3 Apr 2023

1. Improved performance and reduced memory footprint during scoring
2. Enhanced and more robust lock refresh logic
3. Enhanced ambiguous entity processing logic
4. Optimize reevaluations and clustered unresolve
5. Libpostal: updated address parsing data model
6. Add organizational name culture support
7. Improved handling of person names that use or overlap with Chinese characters (Taiwan, Japanese Kanji, etc)
8. Add handling of multiple generational values in names (e.g., John Smith JR JR), and stop removing hyphens from generational values (e.g. John Smith F-JR)
9. Improved name handling of Ukrainian, Russian, and Chinese organizational names
10. Improved compressed name scoring involving TAQs (titles, affixes, qualifiers). TAQs removed in compressed name scoring are still used to adjust scoring (e.g. PENN STATE OIL LLC and PENN STATE OIL CORP are not 100% matches)
11. Add non-ASCII TAQ support
12. Default configuration changes (new configs only):
    • Close FFs (features with a behavior of frequency few, e.g. addresses, phone numbers, etc) create possible relationships. Previously only Same FFs would do so, and Close FFs were returned only on search
    • Close Names without an exclusive denial create a possibly same relationship. Previously these were only returned on search
13. Add apiVersion to workload stats
14. JRE removed from packages
15. Various fixes and improvements

Version: 3.4.2
Released: 9 Feb 2023

1. Fixed: Senzing engine configuration SQL Connection strings now support URL encoded characters. This is useful when passwords contain reserved characters (e.g. ";", "/", "?", ":", "@", "=", and "&"). See https://www.w3schools.com/tags/ref_urlencode.asp for the HTML URL character encoding reference

Version: 3.4.1
Released: 24 Jan 2023

1. Fixed: A rare exception where an in-memory virtual entity is not found, logged as 0087E|SRD Exception 'Failed to find virtual entity in virtual repository.'

Version: 3.4.0
Released: 11 Jan 2023

1. Substantial reduction in DB access and CPU for entity retrieval operations when internal features are not requested (default behavior).
   • NOTE: With one worst-case scenario test of an entity with millions of features, this improvement went from hours to seconds.
   • NOTE: In a 5k record stress test case, over 100k SQL statements were removed.
2. Significant reduction in Distinct Feature calculation CPU via score caching and optimizations
3. How functionality performance improvements
4. Unresolve performance improvements
5. NEW: Added Khmer name culture support
6. Reduce constraint violations that get logged in PostgreSQL application logs
7. Compressed name processing is no longer enabled by default for organization names (e.g., Fish Shack vs FishShack). This change prevents overmatching (especially in Cyrillic org names), and also allows names such as Fish Shack IV vs Fish Shack VI to score properly.
   • NOTE: The behavior on existing deployments won't change unless you use an updated defaultGNRCP.config file.
8. Fixed: Properly threshold ambiguous entities in clustered unresolve
9. Fixed: Invalid RECORD_ID latch contention
10. Note: Having a "lock" stolen has been raised to a logged error for visibility, though the operation will retry. If it happens often, please contact support.
11. Note: The unused RES_FEAT_EKEY table first/last seen date fields are no longer referenced in SQL to improve DB performance.
12. Various fixes and improvements

Version: 3.3.2
Released: 26 Oct 2022

1. Fixed: Enhanced reconnection logic in edge cases where the database server disconnects

Version: 3.3.1
Released: 10 Oct 2022

1. Fixed: A rarely used configuration parameter that prevents exact duplicate records from automatically deduplicating

Version: 3.3.0
Released: 26 Sep 2022

1. Significant performance improvement loading records with many names (e.g., watchlist)
2. Allow multiple redo record processes
3. Confused entities enhancements:
   • Improved and consistent identification of confused entities
   • Large entities no longer considered confused entities
   • Redo and reevaluate now correct confused entities
4. Improve: Add threadStates to workload stats to help more easily identify contention and SQL wait. Note that these are for the instant when G2Engine.stats() was called and NOT some average/cumulative value:

   
     "threadState": {
      "active": 12, # processing a record
      "idle": 0,
      "sqlExecuting": 11, # in sql execute
      "loader": 4, # in the G2 loader, inserting/updating the internal record
      "resolver": 6, # doing ER
      "scoring": 2, # scoring
      "dataLatchContention": 0, # waiting for another thread that shares a similar key (sorted/grouped data)
      "resEntContention": 0 # contention on the same Entity with another thread/process
     }

5. Fix: Remove a potential, unreported, inconsistency in scoring addresses and a few other strings (did not affect names)
6. Fix: Cluster unresolve (rare) may remove valid relationships for entities it doesn’t modify IF it does modify others
7. Fix: Actual duplicate records no longer generate WithInfo AFFECTED_ENTITIES entries
8. Various fixes and improvements

Version: 3.2.0
Released: 25 Aug 2022

1. Substantial overall performance improvement and resource reductions
2. Massively improved single-process multi-threaded performance
   • Simplifies programming with Senzing by reducing the need to use multi-processes to scale
   • Tested beyond 128 threads in a single process
   • G2Loader defaults to 64 threads per process instead of 16
3. Reduction in memory footprint across the application and especially when reducing the number of processes and increasing the number of threads
4. Senzing will now work in Linux/amd64 docker containers run on Apple Silicon processor
5. Fix: deleteRecordWithInfo to now return a WithInfo document populated with the requested DATA_SOURCE and RECORD_ID and the AFFECTED_ENTITIES list empty when the record does not exist
6. Fix: to allow configuration calls to use SoftHSM database security
7. Fix: to handle invalid prepared statements on MSSQL
8. Fix: for zero count entity feature stats. System auto corrects as data is loaded
9. Fix: redo records were classified as "addedRecords" in workload stats
10. Fix: correctly report initial and current "availableMemory" in workload stats
11. Improve: error message when no schema is in the database
12. Improve: move to a private ICU build to prevent conflicts with system install ICU libraries
13. Improve: workload stats now includes total workerThreads and the number of activeThreads processing requests
14. Improve: most databases will have the SYS_EVAL_QUEUE populated with a hashcode for MSG_ID instead of a sequence value to improve the random ordering of redo processing
15. Known Issue: physical cores reported will actually be the same as the number of logical cores
16. Known Issue: when loading or saving a configuration change you may see the following warnings: WARN: Configuration [CFG_DSRC_INTEREST] returned an empty config and WARN: Valid salt found but no FTYPE(s) configured for hashing. These warnings can be ignored
17. >New vastly easier Docker containers that can be used to build custom Senzing-based applications. Available in our Dockerhub and in our GitHub repositories

Version: 3.1.2
Released: 11 Jul 2022

1. Fix to handle invalid cursor errors on MSSQL
2. Improved handling of database connection issues due to database cursor errors
3. Various fixes and improvements

Version: 3.1.1
Released: 22 Jun 2022

1. Fix for RedHat 8 Python dependencies
2. Updates to enable custom plugin functionality
3. Various fixes and improvements

Version: 3.1.0
Released: 7 Jun 2022

1. Fix for zero count entity feature stats. System auto corrects as data is loaded
2. Fix to properly generate and display disclosed relationship labels / MATCH_KEYs. System auto corrects as data is loaded
3. (Python tooling) G2Loader.py fix to allow loading of records where the ENTITY_TYPE is not configured. Records are now loaded as "ENTITY TYPE": "GENERIC"
4. (Python tooling) G2Explorer.py now shows relationships on get calls by default
5. Various fixes and improvements

Version: 3.0.0
Released: 4 May 2022

1. NOTE: THIS RELEASE REQUIRES A CONFIGURATION UPGRADE AND DATABASE SCHEMA UPGRADE. THESE MUST BE PERFORMED
2. IMPORTANT: See Senzing API 3.0 - Breaking Changes
for a detailed list of modified and deprecated features and requirements
3. NEW: howEntityByEntityID() function details how an entity got put together record by record
4. NEW: Oracle database support - Oracle 19c or newer
5. Optimize reevaluations to auto-correct additional data conditions
6. Optimize disclosed relationships
7. Optimize observed features
8. Libpostal address parsing enhancements
   • Improved parsing of Singapore unit information
   • Improved parsing of US unit information
   • Improved parsing of Hawaiian TMK information
   • Improved parsing of Great Britain post codes
   • Improved parsing of post office box data
   • Improved parsing of Spanish and Portuguese unit information
   • Improved parsing of abbreviated and alias words in address data
   • Various libpostal fixes and improvements
9. (Python tooling) G2Explorer.py enhancements
   • new how command tells you how an entity got put together record by record
   • new tree command groups and counts an entity's relationships in a tree view format through degrees of separation
   • Enhanced export command now exports the JSON for an entity and all its relationships through degrees of separation, excellent for creating support tickets and test cases
   • Enhanced set command now allows for toggling relationships to be shown on a get, suppressing other data sources on data source and cross source summary reports, and setting the statistics on those reports to be record based rather than entity based
   • Improved paging through reports now clearly identifies which example number you are on and allows you to jump to any example
10. (Python tooling) G2Loader.py enhancements
    • Enhancement to make workload stats periodic
11. (Python tooling) G2Snapshot.py enhancements
    • new -d parameter to take a snapshot of just a single data source and all it matched. Why take a snapshot of hundreds of millions of records from a reference data set when all you want to know did any match my customers
    • Enhanced performance: G2Snapshot.py is now multi-threaded and restartable.
    • Enhanced performance: start reviewing a snapshot as soon as a chunk of it has been processed, see the -k chunksize parameter
12. Improved handling of database connection issues
13. Fix Linux workload CPU stats
14. Fix to properly handle newlines in JSON
15. Various fixes and improvements

Version: 2.9.1
Released: 3 Aug 2022

1. Fix to handle invalid prepared statements on MSSQL
2. Fix for zero count entity feature stats. System auto corrects as data is loaded
3. Fix to allow configuration calls to use SoftHSM database security
4. Various fixes and improvements

Version: 2.9.0
Released: 28 Jun 2022

1. Fix to handle invalid prepared statements on MSSQL
2. Fix for zero count entity feature stats. System auto corrects as data is loaded
3. Various fixes and improvements

Update: 2.8.8
Released: 31 Mar 2022

1. Fix for selective hashing middle name scoring

Update: 2.8.7
Released: 1 Mar 2022

1. Fix to enhance database reconnects on failed connections

Update: 2.8.5
Released: 12 Jan 2022

1. Enhancement to support a threshold for disclosed relationship suppression for each entity per REL_ANCHOR_DOMAIN or REL_POINTER_DOMAIN value (contact support@senzing.com for enablement instructions)
2. Various fixes and improvements

Update: 2.8.4
Released: 10 Nov 2021

1. Enhancement to support partial SSN handling (contact support@senzing.com for enablement instructions)
2. Various fixes and improvements

Update: 2.8.3
Released: 7 Oct 2021

1. Fix to remove outdated feature values when reloading or reevaluating records

Update: 2.8.2
Released: 7 Sep 2021

1. Fix to prevent match denials due to incomplete or invalid dates of birth under certain data conditions

Update: 2.8.1
Released: 11 Aug 2021

1. Fix to improve incomplete reevaluations of records due to connectivity errors
2. Optimize reevaluations to auto-correct certain data conditions
3. Various fixes and improvements

Version: 2.8.0
Released: 14 Jul 2021

1. Optimize database reconnections on PostgreSQL
2. Enhancement to support Java-language plugins for feature standardization, expression building, and comparison
3. OPTIONAL: You may drop the OBS_FEAT_EKEY_SK index as it is no longer used by the system. This reduces unnecessary IO and storage space
4. Fix to improve handling of non-ASCII characters in addresses
5. Various fixes and improvements
6. (Python tooling) G2Explorer.py enhancements
   • search command now provides additional information for searches to help determine if the entity does not exist, if the search criteria is insufficient, or if there are too many search results
   • Various fixes and improvements

Version: 2.7.0
Released: 14 Jun 2021

1. Fix MSSQL database reconnection
2. Fix NAME_FULL collapsing of initials when scored against NAME_ORG
3. Various fixes and improvements

Version: 2.6.0
Released: 17 May 2021

1. Enhancement to increase performance and reduce IOPS for certain features
2. Enhancement to increase performance and reduce IOPS when processing redo records
3. Fix record locking when processing redo records
4. Various fixes and improvements
5. (Python tooling) G2Command.py enhancements
   • Restart engine when the configuration is modified
   • Improved error messaging
   • Various fixes and improvements
6. (Python tooling) G2ConfigTool.py enhancements
   • get* and delete* commands now autocomplete where relevant. For example, getRule <tab> will list rules
   • get* commands will pretty print JSON by default. Turn off with prettyPrint. See help prettyPrint
   • list* commands now take optional argument to filter results. For example, listRules sf1 will filter only rules that contain sf1 anywhere in the rule
   • list* commands that produce long output lines are now spaced with newlines for easier reading
   • New configuration commands:getActiveConfigID, getConfigList, getConfig, and getConfigSection
   • Various fixes and improvements
   • Fix deleteRule error when using a rule name
7. (Python tooling) G2Loader.py enhancements
   • Improved output format and statistics when running in test mode using -T / --testMode
   • Various fixes and improvements

Update: 2.5.1
Released: 6 May 2021

1. Enhancement to improve PostgreSQL performance when processing redo records
2. Fix to improve handling of non-ASCII characters
3. Various fixes and improvements

Version: 2.5.0
Released: 15 Apr 2021

1. OPTIONAL: Configuration update to implement certain new features and fixes. Click here for instructions
2. Fix PostgreSQL redo queue 40P01 deadlocking error - (requires configuration update)
3. Enhancement to enable usage of Placekey values for address matching - (requires configuration update)
4. Update to the new senzingdata-v2 package. NOTE: If you have a custom install, you must modify it to use the new RPM or APT package names
5. Upgrade Global Name Management libraries and data
6. Use Senzing truthset data as default data if no file or project is specified
7. Fix UTF-8 character conversion in Java API
8. Fix clustering of feature values in output
9. Various address parsing and resolution enhancements
10. Various fixes and improvements
11. (Python tooling) G2Loader.py enhancements
    • Add _-_SzShuf_-_ suffix and timestamp to shuffled files to identify them
    • (command line argument) -snd and --shuffleNoDelete replaces -nsd and --noShuffleDelete to clarify meaning
    • (command line argument) -sfr and --shuffFileRedirect redirects shuffled files to another path to improve disk performance when shuffling large files
    • Pause and prompt to allow a user to exit and use a previously located shuffle file if found, continues without input if used in scripts
    • (command line argument) -sfi and --shuffFilesIgnore ignores check for existing shuffled files
    • Fix compressed file usage with projects
    • Various fixes and improvements

Update: 2.4.2
Released: 7 Apr 2021

1. Fix PostgreSQL redo queue 40P01 deadlocking error - (new installs only, version 2.5.0 includes upgrade fix)

Update: 2.4.1
Released: 10 Mar 2021

1. Fix searchByAttributes and searchByAttributesV2 results when using G2_EXPORT_* flags

Version: 2.4.0
Released: 24 Feb 2021

1. Enhancement for API calls to default ENTITY_TYPE to GENERIC
2. Enhancement to reduce PostgreSQL transaction ID usage
3. Enhancement of new g2dbencrypt tool. Please contact support for usage
4. Enhancement to improve specificity of error codes
5. Enhancement to enable usage of Placekey values for address matching. Contact support@senzing.com for info
6. (Python tooling) G2Export.py enhancements
   • (command line argument) -cf and --compressFile compresses output to a gzip file
   • (command line argument) -x and --extended for CSV now only includes JSON_DATA for the entity not relationships to reduce size
   • (command line argument) -xcr and --extendCSVRelates adds related entities
   • stdout logs/messages now go to g2export.log
   • Fix to prevent command line argument abbreviations
   • Various fixes and improvements
7. (Python tooling) G2Loader.py enhancements
   • (command line argument) -skr and --skipRecords skips the first n records
   • Fix to prevent command line argument abbreviations. Prevent -F use as an abbreviation for --FORCEPURGE
   • Fix PEP8 issues
   • Add check to prevent purge with redo only mode
   • Remove unused modules and imports
   • Remove unused vars
   • Various fixes and improvements
8. Fix to prevent records with different RECORD_TYPE features from merging
9. Fix to delete orphaned OBS_ENTs
10. Various fixes and improvements

Update: 2.3.3
Released: 19 Feb 2021

1. Fix connection pointer is NULL database error when using a specified schema and multithreading

Update: 2.3.2
Released: 10 Feb 2021

1. Fix to preserve the schema name in a PostgreSQL connection during a reconnect

Update: 2.3.1
Released: 26 Jan 2021

1. Fix G2Export fetchNext database access error

Version: 2.3.0
Released: 15 Dec 2020

1. Improve transaction handling for PostgreSQL
2. Fix Requested lookup of ... using unknown value [0]. Value not found errors
3. Fix Conflicting RECORD_ID values issue in addRecordWithReturnedRecordID
4. Various fixes and improvements
5. Enhance G2Command.py
• (command line argument) -ntm and --threadCountMem: maximum memory % to use when calculating threads, default=80
• Output now includes command line arguments used
• Detailed version and configuration information included output
• Various fixes and improvements

Update: 2.2.6
Released: 7 Dec 2020

1. Fix redo records queue query function

Update: 2.2.5
Released: 17 Nov 2020

1. Enhance whyEntities* response with additional disclosed relationship details
2. Fix handling of specialized characters in data. This resolves bad XML parsing errors.

Update: 2.2.4
Released: 2 Nov 2020

1. Enhance reevaluation processing, including improved error reporting
2. Fix reevaluation Unknown DATA_SOURCE value error
3. G2Loader redo only mode (-R) now automatically calculates additional threads unless specified (-nt)

Update: 2.2.3
Released: 29 Oct 2020

1. Improve redo processing

Update: 2.2.2
Released: 23 Oct 2020

1. Improve redo processing
2. Fix Requested lookup of [ETYPE_ID] using unknown value [0]. Value not found error caused by a bad internal hash. The issue is self-correcting, and entity resolution is not affected
3. Fix G2Export fetchNext error
4. Various G2Loader enhancements, including improvements for container usage
5. Various fixes and improvements

Update: 2.2.1
Released: 30 Sep 2020

1. Include missing G2Loader PostgreSQL governor file: governor_postgres_xid.py

Version: 2.2.0
Released: 29 Sep 2020

1. Enhance Python tooling
2. (Python tooling) G2Command.py enhancements
   • purgeRepository command now accepts --FORCEPURGE argument to purge without prompting
   • restart command restarts G2Command and turns off debugging if on, reloads config changes if any made for example using G2ConfigTool
   • restartDebug command restarts G2Command and turns debug on
   • timer command times command execution. Toggles on/off with same command
   • (command line argument) -t and --debugTrace enables debugging
3. (Python tooling) G2CreateProject.py enhancements
   • Improved error handling
   • Add API version to output
4. (Python tooling) G2Export.py enhancements
   • G2Export.py no longer uses G2Project.ini
   • New export completion output formatting
   • Increased CSV field limit to prevent abend on very large JSON_DATA field
   • Fix issue for export row count under certain data conditions
5. (Python tooling) G2Loader.py enhancements
   • G2Loader.py no longer uses G2Project.ini
   • Automatically optimize thread count (-nt command line argument overrides auto-optimization)
   • New startup and statistics format and information: thread calculation, configuration details, file info, shuffling info, redo info, and governor ingestion pause info.
   • (command line argument) -sr and --STOPONRECORD stops loading records at n records
   • (command line argument) -P and --purgeFirst now prompts before purging
   • (command line argument) -FORCEPURGE and --FORCEPURGE purges without prompting
   • (command line argument) -nsd and --noShuffleDelete doesn't delete shuffled files so they can be used again with -ns or --noShuffle arguments
   • New governor class
   • (governor class) Pause G2Loader when PostgreSQL is running out of transaction IDs and prevent ID wrap around
   • (governor class) Automatically use when PostgreSQL is detected as the Senzing repository
   • (governor class) Inform when ingestion is paused and to issue manual vacuum of database
   • (governor class) Ingestion continues when datfrozenxid drops below threshold during a vacuum
   • (governor class) Can be used as a template to write your own governors for G2Loader
6. Improve transliteration and general name handling
7. New addRecordWithReturnedRecordID function
8. Various fixes and improvements

Update: 2.1.1
Released: 22 Sep 2020

1. Various fixes and improvements

Version: 2.1.0
Released: 3 Sep 2020

1. Enhance whyEntityByEntityID output to return info for all entity records
2. Add option for PO Boxes to be exclusive to other addresses if the city/state/postal are unique
3. Fix non-ASCII JSON parsing issues
4. Improve address parsing
5. Various fixes and improvements

Update: 2.0.1
Released: 23 Jul 2020

1. Fix a rare segfault under certain data conditions

Version: 2.0
Released: 16 Jul 2020

1. NOTE: THIS RELEASE REQUIRES AN UPDATE TO YOUR CONFIG AND SCHEMA. Please refer to this article to perform the upgrade
2. Enhanced and simplified Disclosed Relationships. See How to create disclosed relationships
3. Deprecated legacy features, functions, and support. See Senzing API 2.0 - Breaking Changes
4. POC tools now released as the Senzing Exploratory Data Analysis (EDA) tools in the /python directory. The G2Snapshot.py utility takes a snapshot of the current state of the records loaded in a Senzing repository. The G2Explorer.py utility is an interactive command line utility that uses the output from the G2Snapshot.py utility to display interesting statistics and examples of the entities in an Senzing repository at strategic times. The G2Audit.py utility provides a complete analysis of the differences between entity resolution between two snapshots
5. Add the list of records for related entities to getEntity
6. Add why not functionality to Why functions
7. Various fixes and improvements
8. Increase evaluation license limit to 100k records
9. Updated docs.senzing.com

Version: 3.12.1
Released: 29 Oct 2024

1. Upgraded to Senzing API Version 3.12.1 (This version includes all Senzing API updates from version 3.8.0 to 3.12.1)
2. Various enhancements and bug fixes

---------------------------
Version 3.8 releases
released: 20 Nov 2023
end-of-life: 20 Nov 2025
---------------------------