This is a patch release for bug fixes and experimental features.

• Bug Fix

  • Bitmap64 deserilization error - #918
  • Incorrect update count - #947
  • Dependencies lost in pom.xml - #973
  • Error when executing empty batch using PreparedStatement - #977
  • Connection properties may lost - #978
  • ResultSet.getObject(int, Class) may return default value instead of null - #979
  • Redundant classes in shaded jar - #989
  • PreparedStatement.addBatch() timed out - #991
  • Set nullAsDefault to 2 may trigger IllegalArgumentException when handling timestamp with timezone - #999

• New Feature

  • ClickHouseRequestManager is added for customizing query ID and session ID - #870
  • PreparedStatement.getParameterMetaData() is implemented - #972
  • Experimental transaction support - #975
  • Direct file upload and download - #1001
  • Failover across protocols
  • Repeat the same query until timed out, only when session is locked on server - set repeat_on_session_lock to false to turn it off

This is a patch release for bug fixes and experimental features.

• Bug Fix

  • bump gRPC to 1.45.1 - #936
  • fix insert timeout error as mentioned in #943
  • fix issue of parsing slash in connection string - now you don't have to encode slash in query parameters
  • fix JDBC multi-value insert error - #920
  • fix JDBC metadata error on 21.6 and rename_response_column option no longer impact metadata queries - #938

• New Feature

  • clickhouse-cli-client, a wrapper of ClickHouse native command-line, was added mainly for testing
  • load balancing and failover support - #956
  • new JDBC option nullAsDefault for handling null value in both insert and query - see comment at here

This is a patch release mainly for fixing below issues:

• fix issue parsing Tuple with names - #889 (which was not fully fixed in previous release)
• fix unsupported format error when loading data in custom format - #909
• fix precision issue when converting Float and Double to BigDecimal - #921
• fix type conversion issue when dealing with nullable array elements - #924
• fix OOM issue when loading lots of data through http interface - as mentioned in #861
• fix broken query for retrieving client info properties

In addition, a few more changes(mainly for streaming) were merged from 0.3.3 branch, along with some new features:

• new option rename_response_column for removing column prefix
• new option request_chunk_size (defaults to 8192) for streaming - same option will be applied to gRPC client in the near future
• new options request_buffering and response_buffering for customizing buffering mode - more details will be covered in #928
  • Resource-Efficient the default mode which consumes less CPU and memory
  • Performance uses more CPU and much more memory for better performance
  • Custom allows you to customize options like buffer_size, max_buffer_queue, buffer_queue_variation, use_blocking_queue etc. to balance performance and resource utilization

This is a patch release mainly for fixing below issues:

• fix issue with null in legacy JDBC driver json response deserializer - #879
• fix issue set String as Enum value in new JDBC driver - #883
• fix issue when deserializing array of DateTime with timezone - #886
  Note: for example Map(String,Array(Nullable(DateTime64(3,'Asia/Shanghai'))))
• fix key type when deserializing map objects - #887
• fix issue parsing Tuple with names - #889

In addition, more changes(mostly related to streaming) were merged from 0.3.3 branch along with some new features:

• TLS authentication support - #523
• support SimpleAggregateFunction data type
• initial support of experimental data type JSON/Object('JSON')

  select tuple(1,[2,3], map('4', 5))::Tuple(Int32, Array(Int32), Map(LowCardinality(String), Int32))
  select tuple(1,[2,3], map('4', 5))::Tuple(i1 Int32, a1 Array(Int32), m1 Map(LowCardinality(String), Int32))

If you're using Java client, please be aware of below changes which may or may not cause compiling errors:

• many methods changed to return ClickHouseInputStream/ClickHouseOutputStream instead of plain InputStream/OutputStream
• ClickHouseLZ4InputStream and ClickHouseLZ4OutputStream are removed

This is a patch release for fixing below issues in both Java client and JDBC driver.

• fix backward compatibility issue in legacy driver - #862
• fix rounding error when converting Float/Double value to BigDecimal - #864
• enable read-only support in JDBC driver - #866

In addition, there are a few more changes were merged from 0.3.3 branch:

• ResultSet.unwrap() now supports ClickHouseRecord(current row) in addition to ClickHouseResponse(whole results)
• ClickHouseColumn now provides estimated byte length, which can be used in read/write optimization
• synchronous methods ClickHouseClient.executeAndWait() and Mutation.sendAndWait() for ease of use

This is a patch release for fixing below issues in both Java client and JDBC driver.

• enhance Bool support - #848
• fix confusing error message metioned in #849
• fix issue of writing Date32 value in RowBinary format
• fix OffsetDateTime parsing error
• fix NumberFormatExcption when parsing primitive array with null value
• fix incorrect column type used in Connection.createArrayOf method
• enhance PreparedStatement by treating insert into table(col1,col2,col3) values(?,?,?) as insert into table(col1,col2,col3)

This is a patch release primarily for fixing below issues in JDBC driver. You don't have to upgrade if you're using Java client.

• continueBatchOnError was ignored in Statement.executeBatch()
• executeQuery() may return null
• PreparedStatement.executeUpdate() did not work for insert statement using input function
• Exception was swallowed in Statement.executeUpdate() - #836

In addition, exception handling is improved along with some minor changes:

• support largeMaxRows, largeUpdateCount and executeLargeBatch
• disable execute(*), executeQuery(*), execute*Update(*), and addBatch(*) methods in PreparedStatement to follow JDBC standard
• throw BatchUpdateException instead of SQLException in execute*Batch method
• empty batch is no longer allowed - addBatch is a required before execute*Batch
• query in batch is not allowed, according to JDBC standard
• mixed use of addBatch and execute / executeQuery / execute*Update is not allowed
• ported some IO classes from 0.3.3 branch for slightly improved performance

This is a patch release for fixing issues in both Java client and JDBC driver:

• Incorrect value may return when converting large UInt64 value from String to Long - #828
• Not able to create JDBC connection when database(defined in connection property or url) does not exist - #826
  Note: createDatabaseIfNotExist connection property was added as suggested in #605.

This is a patch release primarily for fixing below issues in JDBC driver:

• Incorrect COLUMN_SIZE, DECIMAL_DIGITS, and CHAR_OCTET_LENGTH were returned from ClickHouseDatabaseMetaData.getColumns() - #807
• Failed to parse insert statement with CTE - #810
• PreparedStatement.setBytes() and ResultSet.getBytes() do not work

In addition, please be aware of below changes related to ClickHouseDataType:

• Enum is deprecated and it will be removed in 0.3.3 - it should be an alias of Enum8
• More aliases like VARBINARY were added

This is a patch release for fixing issue of getting table meta data from ClickHouse server older than 21.6 - see #804 for details.

Java client does not have the issue, so you can continue to use 0.3.2.